Method for constructing retail credit risk prediction model and credit card service Scorealpha model

Abstract

The invention relates to a credit card service Scorealpha model and a method for predicting retail credit risks by using the model, and the method comprises the steps: a data collection step: obtaining retail credit prediction data of a to-be-predicted sample; a step of classifying the to-be-predicted samples, wherein the to-be-predicted samples are classified based on a decision tree method so as to determine a sub-model used for calculating the credit default probability; and a credit default probability calculation step of substituting the retail credit prediction data into the credit default probability sub-model to calculate the credit default probability of the to-be-predicted sample. The invention also relates to a method for constructing the retail credit risk prediction model. The method comprises a data acquisition step, a data derivation step, a feature preliminary screening step, a preliminary screening data conversion step, a feature fine screening step and a credit default probability modeling step.

Description

Technical Field

[0001] This application relates to a credit risk management system and method. The method and system of this application can assist financial institutions in making more accurate risk decisions and accelerate their digital transformation process. In detail, this application relates to a method for retail credit risk prediction model and a credit card business Scorealphad model. Background Art

[0002] Credit cards are the most important credit business of banks. The manual approval mechanism of some financial institutions can no longer cope with the increasing credit demand, so there is an urgent need to improve the intelligent risk control capabilities of financial institutions. Financial institutions hope to build a risk mitigation mechanism for the entire credit process, from customer pre-screening, pre-loan review, loan approval, post-loan management to the early collection stage.

[0003] If a scoring system can be developed based on the principles of early identification, early warning, early detection and early disposal, and credit business can be monitored and managed quickly and conveniently, the financial institutions' own business volume, competitive advantage and asset quality can be improved while risks are controllable.

[0004] However, the construction of a scoring system is highly dependent on data and technology. The diversity and coverage of data dimensions, modeling techniques and methodology directly affect the final stability and ranking of the scoring system. Some financial institutions have little experience in intelligent risk control of retail business and have weak risk control capabilities. In the actual application process, factors such as lack of data mining and analysis capabilities and weak risk modeling technology have led to financial institutions being unable to fully utilize the value of internal data and effectively improve the accuracy and stability of models, and other control and management technical difficulties. This is also one of the main obstacles faced by small and medium-sized financial institutions in the direction of digital transformation. Summary of the invention

[0005] In view of the deficiencies in the above-mentioned prior art, this application is intended to provide a credit risk management system and method, which can provide effective risk management for financial institutions. The credit risk prediction method and system of this application, namely the method of retail credit risk prediction model and the credit card business Scorealphad model, are based on the massive data of large banks and the risk laws extracted by statistical principles, which are essentially of promotional significance.

[0006] This application is applicable to credit risk management in credit business, especially for credit card business scenarios. Credit card business is the most important credit business of commercial banks, which can create huge profits and income for banks, and is also the most convenient and frequently used credit product by retail customers.

[0007] In order to avoid credit risk, a relatively objective evaluation of the borrower's creditworthiness should be made when granting credit services, for example, a credit scoring method can be used. The construction process of other popular scoring models on the market is often limited by unfavorable factors such as a small number of modeling samples, a relatively single data source, and high homogeneity of data dimensions. At the same time, since most of the risk assessment models on the market currently use data with weak financial attributes, that is, they are based on non-credit transaction data such as smart terminal device data, social platform data, online shopping mall data, and non-overdue prediction targets for modeling, their prediction results often have a large deviation from the actual credit overdue situation.

[0008] The method and system of this application are developed based on highly stable and high-coverage data samples, and systematically innovate the relatively mature credit risk control system. Credit samples containing various business forms are used to predict the potential credit risks of financial institutions in credit card business. Specifically, this application uses credit card customer sample modeling to mine the credit and asset characteristics of credit card customers and construct a credit scoring method that is more suitable for commercial banks with credit cards as their main business type.

[0009] Specifically, the present application can further predict the probability of credit default in the future based on (1) the customer group whose account age for applying for credit business is less than G1 (for example, G1 is 6); (2) the customer group whose account age for applying for credit business is greater than or equal to G1 and is currently overdue; (3) the customer group whose account age for applying for credit business is greater than G1, who is not currently overdue and whose current credit limit utilization rate is greater than or equal to a%; (4) the customer group whose account age for applying for credit business is greater than G1, who is not currently overdue and whose current credit limit utilization rate is less than a% (for example, a is 100); (5) the customer group who has not applied for credit business and whose account age is less than G2 (for example, G2 is 3) predicts the probability of credit default; these five credit default probabilities are used as prediction targets (i.e., target variables) for research and development.

[0010] Compared with existing models on the market, the method and system of the present application have greatly improved the discrimination and stability in predicting credit delinquency of more than 90 days for this specific customer group.

[0011] This application involves the following technical solutions:

[0012] 1. A method for calculating retail credit risk, comprising:

[0013] A data collection step, which obtains retail credit prediction data of the sample to be predicted;

[0014] A step of classifying samples to be predicted, which classifies the samples to be predicted based on a decision tree method to determine a sub-model for calculating the probability of credit default;

[0015] The credit default probability calculation step is to substitute the retail credit prediction data into the credit default probability sub-model to calculate the credit default probability of the sample to be predicted, and preferably the credit default probability is the credit default probability.

[0016] 2. The method according to claim 1, further comprising:

[0017] After calculating the credit default probability, a step of calculating the credit score of the sample to be predicted is used to calibrate the calculated credit default probability to a standardized score of 0-1000 points.

[0018] 3. The method according to item 1 or 2, wherein:

[0019] The retail credit prediction data includes original retail credit prediction data of the sample to be predicted and derived retail credit prediction data processed based on the original retail credit prediction data;

[0020] Preferably, the original retail credit prediction data includes:

[0021] Credit card basic data, which is based on all available data during the sample user's credit card creation and usage process.

[0022] Basic data on personal loans, which is all available data based on the loan application and usage behavior of sample users.

[0023] Basic customer information data, which is based on the attributes of the sample users themselves but is not directly related to their behavior in financial institutions.

[0024] Basic data on personal financial assets, which includes all other financial assets and financial transaction data of sample users in financial institutions that are not related to credit cards and loans.

[0025] 4. The method according to any one of items 1 to 3, wherein

[0026] The derived retail credit prediction data processed based on the original retail credit prediction data refers to the data obtained by processing the collected original retail credit prediction data based on the time dimension, space dimension, frequency dimension and statistical information dimension;

[0027] Preferably, the derived retail credit forecast data includes but is not limited to:

[0028] Derived retail credit forecast data obtained by processing based on sample relationship length,

[0029] Derivative retail credit forecast data obtained by processing time interval variables,

[0030] Derivative retail credit forecast data obtained by processing based on the frequency of sample behavior,

[0031] Derivative retail credit forecast data obtained by processing the current time point of the sample,

[0032] Derivative retail credit forecast data obtained by processing based on sample continuous behavior,

[0033] Derived retail credit forecast data obtained by processing sample data based on statistical information dimensions.

[0034] 5. The method according to any one of items 1 to 4, wherein

[0035] The retail credit prediction data is selected from one, two, three, four, five, six or seven of the following:

[0036] The current monthly credit repayment rate, the current remaining available credit limit of the credit card, the number of months since the credit card credit limit utilization rate in the past 12 months was >90%, the average number of cash withdrawals from the credit card in the past 3 months, the maximum asset size value in the past 3 months, the number of months since the minimum balance of the deposit account in the past 12 months, the number of months in which the credit repayment rate in the past 3 months was >80%, the number of credit overdue payments >0 in the past 6 months, the current credit card cash withdrawal balance (including cash installments) / current credit limit percentage, the current monthly credit limit utilization rate of revolving loans, the corresponding coefficient of the number of months with interest >0 for special installments in the past 12 months, the number of consecutive months with the largest month-on-month increase in asset size in the past 3 months, the percentage of overdue interest on mortgage loans in the past 6 months to all repayment amounts in the past 6 months, the proportion of months with credit limit utilization rate >70% in the past 3 months, the past 6 Minimum monthly credit card repayment rate, number of months since the credit card interest rate was >0 in the past 3 months, percentage of total cash withdrawal amount of credit card consumption in the current month (including installments) / total cash withdrawal amount of consumption in the previous month (including installments), current deposit balance, average balance of investment and financial management accounts in the past 12 months, percentage of the largest consecutive increase in asset size in the past 3 months, current remaining available credit card limit, percentage of months with credit card repayment rate <90% in the past 12 months, number of months with the largest consecutive increase in credit card consumption balance (including consumption installments) in the past 3 months, number of months of credit card revolving and installment use in the past 6 months, average monthly asset size in the past 6 months / average monthly asset size in the past 12 months, maximum salary in the past 12 months, maximum balance of investment and financial management accounts in the past 6 months, minimum balance of deposit accounts in the past 6 months.

[0037] 6. The method according to any one of items 1 to 5, wherein

[0038] The step of classifying the sample to be predicted includes the following sub-steps:

[0039] Whether the sample to be tested is a customer who has applied for credit business from a financial institution;

[0040] Whether the sample to be tested has an age greater than or equal to G1;

[0041] Whether the sample to be tested is a customer with overdue credit;

[0042] Whether the current credit utilization rate of the sample to be tested is less than a%;

[0043] Based on the above sub-steps, the samples to be predicted are classified to determine the sub-model for calculating the probability of credit default. Under the premise of ensuring the rationality of business logic, the order of the above sub-steps can be set arbitrarily;

[0044] It is preferred to classify the predicted samples in the following order:

[0045] First, determine whether the sample to be tested is a customer who has applied for credit business at a financial institution;

[0046] Then determine whether the sample to be tested has an account age greater than or equal to G1;

[0047] Then determine whether the credit of the sample to be tested is overdue;

[0048] Then determine whether the current credit limit utilization rate of the sample to be tested is less than a%.

[0049] 7. The method according to any one of items 1 to 6, wherein

[0050] After performing feature conversion on the retail credit prediction data, the data is substituted into the credit default probability sub-model to calculate the credit default probability of the sample to be predicted. The feature conversion step includes:

[0051] Based on the feature type of the retail credit prediction data that needs to be substituted into the credit default probability sub-model, the WOE method, dummy feature method or continuous method is selected for feature conversion.

[0052] 8. The method according to claim 7, wherein:

[0053] The continuous method for feature conversion includes the following methods: directly selecting the original value, calculating the square of the original data, calculating the square root of the original data, calculating the cube root of the original data, or calculating the natural logarithm of the original data to perform continuous feature conversion.

[0054] 9. The method according to any one of items 1 to 8, wherein

[0055] The credit default probability sub-model is a model constructed based on sample retail credit prediction data and credit default probability using logistic regression based on the existing user population.

[0056] 10. The method according to any one of items 1 to 9, wherein

[0057] The retail credit forecast data is selected from: the current monthly credit repayment rate, the current remaining available credit limit of the credit card, the number of months since the credit card credit limit utilization rate > 90% in the past 12 months, the average number of cash withdrawals from the credit card in the past three months, and one, two, three, four or five of the maximum asset size values ​​in the past three months.

[0058] 11. The method according to item 10, wherein the step of calculating the probability of credit default comprises:

[0059] The retail credit prediction data is selected from: the current monthly credit repayment rate, the current remaining available credit limit of the credit card, the number of months since the credit limit utilization rate of the credit card in the past 12 months was > 90%, the average number of cash withdrawals from the credit card in the past three months, and the maximum value of the asset size in the past three months;

[0060] Preferably,

[0061] The current monthly repayment rate of the credit card is converted in a continuous manner; the current remaining available credit limit of the credit card is converted in a continuous manner; the number of months since the credit limit utilization rate of the credit card in the past 12 months was >90% is converted in a continuous manner; the average number of cash withdrawals from the credit card in the past three months is converted in a WOE manner; the maximum value of the asset size in the past three months is converted in a continuous manner;

[0062] More preferably,

[0063] The current monthly credit repayment rate is converted into the original value calculation method using a continuous method; the current remaining available credit limit of the credit card is converted into the natural logarithm calculation method using a continuous method; the number of months from now when the credit limit utilization rate of the credit card in the past 12 months is >90% is converted into the cube root calculation method using a continuous method; the maximum asset size value in the past 3 months is converted into the square root calculation method using a continuous method.

[0064] 12. The method according to claim 11, wherein:

[0065] The converted values ​​of the five features, namely, the current monthly credit repayment rate, the current remaining available credit limit of the credit card, the number of months since the credit card limit utilization rate > 90% in the past 12 months, the average number of cash withdrawals from the credit card in the past three months, and the maximum asset size value in the past three months, are substituted into the sub-model constructed by logistic regression based on the sample retail credit prediction data and credit default probability to calculate the default probability of the sample to be predicted.

[0066] 13. The method according to claim 12, wherein:

[0067] The sub-model is shown in the following formula 1:

[0068]

[0069] Where k is the number of features entering the model, preferably k is 5,

[0070] α is the intercept term, the value range is (-0.58156, -0.82194), and the optimal value is -0.701748;

[0071] β1 is the coefficient corresponding to the current monthly loan repayment rate, with a value range of (-0.68379, -0.78681), and the optimal value is -0.7353;

[0072] β2 is the coefficient corresponding to the current remaining available credit limit of the credit card, with a value range of (-0.60093, -0.66961), and the optimal value is -0.635274;

[0073] β3 is the coefficient of the number of months from now when the credit limit utilization rate of the credit card in the past 12 months is >90%, with a value range of (-0.29231, -0.60137) and the optimal value is -0.446841;

[0074] β4 is the coefficient corresponding to the average number of cash withdrawals of credit cards in the past three months, with a value range of (-0.6339, -0.73691) and an optimal value of -0.685404;

[0075] β5 is the coefficient corresponding to the maximum value of asset size in the past three months, with a value range of (-0.08642, -0.22378) and an optimal value of -0.155097;

[0076] x1 is the original value of the current monthly credit repayment rate generated by the feature conversion step;

[0077] x2 is the natural logarithm transformation value of the current remaining available credit limit of the credit card generated in the feature conversion step;

[0078] x3 is the cube root conversion value of the number of months from now when the credit card limit utilization rate in the past 12 months is >90% generated in the feature conversion step;

[0079] x4 is the WOE conversion value of the average number of cash withdrawals from credit cards in the past three months generated in the feature conversion step;

[0080] x5 is the square root transformed value of the maximum asset size value in the past three months generated in the feature transformation step.

[0081] 14. The method according to claim 13, wherein:

[0082] After calculating the credit default probability, the step of calculating the credit score of the sample to be predicted is to use the following formula to calculate and generate a credit score for characterizing the borrower:

[0083]

[0084]

[0085] Among them, P is the borrower's default probability (P) generated in the credit default probability calculation module, A is 54.2458; B is 115.4156, and the round function rounds the calculated score to the integer value; finally, the score greater than 1000 is set to 1000 points, and the score less than 0 is set to 0 points.

[0086] 15. The method according to any one of items 1 to 9, wherein

[0087] The retail credit forecast data is selected from: the number of months from the minimum balance of the deposit account in the past 12 months, the number of months with a credit repayment rate > 80% in the past 3 months, the number of credit overdue payments > 0 in the past 6 months, the current cash withdrawal balance of the credit card (including cash installments) / current credit limit percentage, the current monthly credit limit utilization rate of revolving loans, and one, two, three, four, five or six corresponding coefficients of the number of months with special installment interest > 0 in the past 12 months.

[0088] 16. The method according to item 15, wherein the step of calculating the probability of credit default comprises:

[0089] The number of months from the minimum balance of the deposit account in the past 12 months, the number of months with a credit repayment rate > 80% in the past 3 months, the number of credit overdue payments > 0 in the past 6 months, the current credit card withdrawal balance (including cash installments) / current credit limit percentage, the current monthly credit limit utilization rate of revolving loans, and the number of months with interest > 0 in the past 12 months for special installments are converted into features.

[0090] Preferably,

[0091] The number of months from the minimum balance of the deposit account in the past 12 months to the present is converted in a continuous manner; the number of months with a credit repayment rate > 80% in the past 3 months is converted in a continuous manner; the number of credit overdue payments > 0 in the past 6 months is converted in a continuous manner; the current credit card cash withdrawal balance (including cash installments) / current credit limit percentage is converted in a continuous manner; the current monthly credit limit utilization rate of revolving loans is converted in a continuous manner; the number of months with interest > 0 in the past 12 months of special installments is converted in a continuous manner;

[0092] More preferably,

[0093] The number of months from the minimum balance of the deposit account in the past 12 months to the present is converted in a continuous manner to the number of months from the present minimum balance of the deposit account in the past 12 months using the cube root calculation method; the number of months in which the credit repayment rate in the past 3 months was >80% is converted in a continuous manner to the number of months in which the credit repayment rate in the past 3 months was >80% using the original value calculation method; the number of credit overdue payments >0 in the past 6 months is converted in a continuous manner to the number of credit overdue payments >0 in the past 6 months using the original value calculation method; the current cash withdrawal balance (including cash installments) of the credit card / current credit limit percentage is converted in a continuous manner to the current credit card cash withdrawal balance (including cash installments) / current credit limit percentage using the natural logarithm calculation method; the current monthly credit limit utilization rate of the revolving loan is converted in a continuous manner to the original value calculation method; the number of months in which the interest on the special installment loan is >0 in the past 12 months is converted in a continuous manner to the number of months in which the interest on the special installment loan is >0 in the past 12 months using the square root calculation method.

[0094] 17. The method according to claim 15, wherein:

[0095] The converted values ​​of the six features, namely, the number of months from the present when the minimum balance of the deposit account in the past 12 months was from now, the number of months with a credit repayment rate > 80% in the past 3 months, the number of credit overdue payments > 0 in the past 6 months, the current cash withdrawal balance of the credit card (including cash installments) / current credit limit percentage, the current monthly credit limit utilization rate of revolving loans, and the number of months with special installment interest > 0 in the past 12 months, are substituted into the sub-model constructed by logistic regression based on the sample retail credit prediction data to calculate the default probability of the sample to be predicted.

[0096] 18. The method according to claim 17, wherein:

[0097] The sub-model is shown in the following formula 2:

[0098]

[0099] Where k is the number of features entering the model, and k is preferably 6;

[0100] α is the intercept term, the value range is (-0.14511,-0.18322), and the optimal value is -0.164164;

[0101] β1 is the coefficient corresponding to the number of months from the present when the minimum balance of the deposit account was the same in the past 12 months. The value range is (0.00429, -0.06312), and the optimal value is -0.029414;

[0102] β2 is the coefficient corresponding to the number of months in which the loan repayment rate was >80% in the past three months, with a value range of (0.26668, 0.12932) and an optimal value of 0.198;

[0103] β3 is the coefficient corresponding to the number of times the credit is overdue > 0 in the past 6 months, with a value range of (-0.48885, -0.55331) and an optimal value of -0.52108;

[0104] β4 is the coefficient corresponding to the current cash withdrawal balance (including cash installment) of the credit card / current credit limit percentage, with a value range of (-0.51312, -0.51864), and the optimal value is -0.515879;

[0105] β5 is the coefficient corresponding to the utilization rate of the current monthly quota of the revolving loan, with a value range of (-0.32363, -0.35507), and the optimal value is -0.339353;

[0106] β6 is the coefficient corresponding to the number of months with interest > 0 in the past 12 months of the special installment plan, with a value range of (-0.38218, -0.63469) and an optimal value of -0.508438;

[0107] x1 is the cube root conversion value of the minimum balance of the deposit account in the past 12 months generated by the feature conversion step.

[0108] x2 is the original value of the number of months with a credit repayment rate > 80% in the past three months generated by the feature conversion step;

[0109] x3 is the original value of the number of credit overdue > 0 in the past 6 months generated by the feature conversion step;

[0110] x4 is the natural logarithm transformation value of the current cash withdrawal balance (including cash installments) of the credit card / current credit limit percentage generated in the feature conversion step;

[0111] x5 is the original value of the current monthly credit limit utilization rate of the revolving loan generated in the feature conversion step;

[0112] x6 is the square root conversion value of the number of months with interest > 0 in the past 12 months for the special installment generated in the feature conversion step.

[0113] 19. The method according to claim 18, wherein:

[0114] After calculating the credit default probability, the step of calculating the credit score of the sample to be predicted is to use the following formula to calculate and generate a credit score for characterizing the borrower:

[0115]

[0116]

[0117] Among them, P is the borrower's default probability (P) generated in the credit default probability calculation module, A is 54.2458; B is 115.4156, and the round function rounds the calculated score to the integer value; finally, the score greater than 1000 is set to 1000 points, and the score less than 0 is set to 0 points.

[0118] 20. The method according to any one of items 1 to 9, wherein

[0119] The retail credit forecast data is selected from: the number of consecutive months with the largest month-on-month increase in asset size in the past three months, the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months, the percentage of months with credit limit utilization rate > 70% in the past three months, the minimum credit card repayment rate in the past six months, the number of months since the credit card interest rate > 0 in the past three months, and the percentage of total cash withdrawal amount of credit card consumption in the current month (including installments) / total cash withdrawal amount of consumption in the previous month (including installments).

[0120] 21. The method according to claim 20, wherein the step of calculating the probability of credit default comprises:

[0121] The features are converted into the maximum number of consecutive months of asset size increase month-on-month in the past three months, the percentage of overdue interest on mortgage loans in the past six months to all repayments in the past six months, the percentage of months with credit limit utilization rate > 70% in the past three months, the minimum credit card repayment rate in the past six months, the number of months since the credit card interest rate > 0 in the past three months, and the percentage of total cash withdrawal amount of credit card consumption in the current month (including installments) / total cash withdrawal amount of consumption in the previous month (including installments).

[0122] Preferably,

[0123] The continuous method is used to convert the number of months with the largest continuous increase in the asset size value month-on-month in the past three months; the continuous method is used to convert the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months; the continuous method is used to convert the proportion of months with credit limit utilization rate > 70% in the past three months; the dummy variable method is used to convert the minimum credit card repayment rate in the past six months; the continuous method is used to convert the number of months since the credit card interest rate > 0 in the past three months; the WOE method is used to convert the percentage of the total amount of cash withdrawals for credit card consumption in the current month (including installments) / the total amount of cash withdrawals for consumption in the previous month (including installments);

[0124] More preferably,

[0125] The continuous method is used to convert the number of months with the largest consecutive month-on-month increase in asset size over the past three months into the number of months with the largest consecutive month-on-month increase in asset size over the past three months using the natural logarithm calculation method; the continuous method is used to convert the percentage of overdue interest on mortgage loans in the past six months to the total repayment amount in the past six months into the percentage of overdue interest on mortgage loans in the past six months to the total repayment amount in the past six months using the original value calculation method; the continuous method is used to convert the proportion of months with credit limit utilization rate > 70% in the past three months into the proportion of months with credit limit utilization rate > 70% in the past three months using the cube root calculation method; the continuous method is used to convert the number of months from now when the credit card interest rate in the past three months is >0 into the number of months from now when the credit card interest rate in the past three months is >0 using the natural logarithm calculation method.

[0126] 22. The method according to claim 20, wherein:

[0127] The converted values ​​of six features, including the number of months with the largest month-on-month increase in asset size in the past three months, the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months, the percentage of months with credit limit utilization rate > 70% in the past three months, the minimum credit card repayment rate in the past six months, the number of months since the credit card interest rate > 0 in the past three months, and the percentage of total cash withdrawals for credit card consumption in the current month (including installments) / total cash withdrawals for consumption in the previous month (including installments), are substituted into the sub-model constructed by logistic regression based on the sample retail credit prediction data and credit default probability to calculate the default probability of the sample to be predicted.

[0128] 23. The method according to claim 22, wherein:

[0129] The sub-model is shown in Formula 3 below:

[0130]

[0131] Where k is the number of features entering the model, and k is preferably 6;

[0132] α is the intercept term, the value range is (-0.11402, -0.35754), and the optimal value is -0.2357787;

[0133] β1 is the coefficient corresponding to the maximum number of consecutive months of increase in asset size in the past three months, with a value range of (-0.19897, -0.31636) and an optimal value of -0.2576679;

[0134] β2 is the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months, with a value range of (-0.03843, -0.28001), and the optimal value is -0.1592187;

[0135] β3 is the coefficient corresponding to the proportion of months with a credit limit utilization rate of >70% in the past three months, with a value range of (-0.71885, -0.91682) and an optimal value of -0.8178348;

[0136] β4 is the coefficient corresponding to the minimum credit card repayment rate in the past 6 months, with a value range of (-0.18116, -0.49436) and an optimal value of -0.3377601;

[0137] β5 is the coefficient corresponding to the number of months from now when the credit card interest rate in the past three months is greater than 0, with a value range of (-0.22644, -0.47791), and the optimal value is -0.352176;

[0138] β6 is the percentage coefficient of the total amount of cash withdrawals from credit card consumption in the current month (including installments) / the total amount of cash withdrawals from consumption in the previous month (including installments). The value range is (-0.05659, -0.38296), and the optimal value is -0.2197794;

[0139] x1 is the natural logarithm-transformed value of the maximum number of consecutive months of increase in asset size over the past three months generated in the feature conversion step;

[0140] x2 is the original value of the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months generated by the feature conversion step;

[0141] x3 is the cube root conversion value of the percentage of months with a credit limit utilization rate > 70% in the past three months generated by the feature conversion step;

[0142] x4 is the dummy variable conversion value of the minimum credit card repayment rate in the past 6 months generated in the feature conversion step;

[0143] x5 is the natural logarithm transformation value of the number of months from now when the credit card interest rate in the past three months is >0, generated in the feature conversion step;

[0144] x6 is a feature conversion step that generates a WOE conversion value of the percentage of the total cash withdrawal amount (including installments) of current month's credit card consumption / total cash withdrawal amount (including installments) of last month's consumption.

[0145] 24. The method according to claim 23, wherein:

[0146] After calculating the credit default probability, the step of calculating the credit score of the sample to be predicted is to use the following formula to calculate and generate a credit score for characterizing the borrower:

[0147]

[0148]

[0149] Among them, P is the borrower's default probability (P) generated in the credit default probability calculation module, A is 54.2458; B is 115.4156, and the round function rounds the calculated score to the integer value; finally, the score greater than 1000 is set to 1000 points, and the score less than 0 is set to 0 points.

[0150] 25. The method according to any one of items 1 to 9, wherein

[0151] The retail credit forecast data is selected from: the deposit balance at the current point in time, the average balance of investment and financial management accounts in the past 12 months, the proportion of the largest number of consecutive months of asset size increase in the past three months, the current remaining available credit card limit, the proportion of months with credit card repayment rate < 90% in the past 12 months, the largest number of consecutive months of increase in credit card consumption balance (including consumption installments) in the past three months, and one, two, three, four, five, six or seven of the number of months of credit card revolving and installment use in the past six months.

[0152] 26. The method according to item 25, wherein the step of calculating the probability of credit default comprises:

[0153] The current deposit balance, the average balance of investment and financial management accounts in the past 12 months, the proportion of the largest consecutive increase in asset size in the past three months, the current remaining available credit card limit, the proportion of months with a credit card repayment rate < 90% in the past 12 months, the largest consecutive increase in credit card consumption balance (including consumption installments) in the past three months, and the number of months of credit card cycle and installment use in the past six months are converted into features.

[0154] Preferably,

[0155] The current deposit balance is converted using a continuous method; the average balance of investment and financial management accounts in the past 12 months is converted using a continuous method; the proportion of the largest consecutive increase in asset size in the past three months is converted using a continuous method; the current remaining available credit card amount is converted using a continuous method; the proportion of the number of months with a credit card repayment rate <90% in the past 12 months is converted using a continuous method; the maximum number of consecutive increase in credit card consumption balances (including consumption installments) in the past three months is converted using a dummy variable method; the number of months of credit card cycle and installment use in the past six months is converted using a continuous method;

[0156] More preferably,

[0157] The current deposit balance is converted using a continuous method to a calculation method using the natural logarithm of the current deposit balance; the average balance of the investment and financial management account in the past 12 months is converted using a continuous method to a calculation method using the natural logarithm of the average balance of the investment and financial management account in the past 12 months; the proportion of the largest consecutive increase in the asset size value in the past 3 months is converted using a continuous method to a calculation method using the original value; the current remaining available credit limit of the credit card is converted using a continuous method to a calculation method using the square root of the current remaining available credit limit; the proportion of the number of months with a credit card repayment rate of <90% in the past 12 months is converted using a continuous method to a calculation method using the original value of the proportion of the number of months with a credit card repayment rate of <90% in the past 12 months; the number of months of credit card revolving and installment use in the past 6 months is converted using a continuous method to a calculation method using the cube root of the number of months of credit card revolving and installment use in the past 6 months.

[0158] 27. The method according to claim 25, wherein:

[0159] The converted values ​​of the seven features, namely, the current deposit balance, the average balance of investment and wealth management accounts in the past 12 months, the proportion of months with the largest consecutive increase in asset size in the past three months, the current remaining available credit card limit, the proportion of months with credit card repayment rate < 90% in the past 12 months, the maximum number of consecutive months with credit card consumption balance (including consumption installments) in the past three months, and the number of months of credit card revolving and installment use in the past six months, are substituted into the sub-model constructed by logistic regression based on the sample retail credit prediction data and credit default probability to calculate the default probability of the sample to be predicted.

[0160] 28. The method according to claim 27, wherein:

[0161] The sub-model is shown in Formula 4 below:

[0162]

[0163] Where k is the number of features entering the model, and k is preferably 7;

[0164] α is the intercept term, the value range is (-0.12673,-0.19974), and the optimal value is -0.1632328;

[0165] β1 is the coefficient corresponding to the deposit balance at the current time point, with a value range of (-0.44362, -0.53469), and the optimal value is -0.4891512;

[0166] β2 is the coefficient corresponding to the average balance of investment and financial management accounts in the past 12 months, with a value range of (-0.181, -0.52118) and an optimal value of -0.3510896;

[0167] β3 is the coefficient corresponding to the proportion of the largest consecutive increase in asset size in the past three months, with a value range of (0.77324, 0.69876) and an optimal value of 0.736;

[0168] β4 is the coefficient corresponding to the current remaining available credit limit of the credit card, with a value range of (-0.16337, -0.27663), and the optimal value is -0.22;

[0169] β5 is the coefficient corresponding to the proportion of months with credit card repayment rate < 90% in the past 12 months, with a value range of (0.1674, 0.0106) and an optimal value of 0.089;

[0170] β6 is the coefficient corresponding to the maximum number of consecutive months of increase in credit card consumption balance (including consumption installments) in the past three months, with a value range of (-0.57383, -0.78447) and an optimal value of -0.6791512;

[0171] β7 is the coefficient corresponding to the number of months of credit card cycle and installment use in the past 6 months, with a value range of (-0.01967, -0.06671) and an optimal value of -0.0431908;

[0172] x1 is the natural logarithm transformation value of the deposit balance at the current time point generated by the feature transformation step;

[0173] x2 is the natural logarithm transformation value of the average balance of investment and financial management accounts in the past 12 months generated in the feature transformation step;

[0174] x3 is the original value of the proportion of the largest consecutive increase in asset size in the past three months generated by the feature conversion step;

[0175] x4 is the square root conversion value of the current remaining available credit limit of the credit card generated in the feature conversion step;

[0176] x5 is the original value of the percentage of months with a credit card repayment rate < 90% in the past 12 months generated by the feature conversion step;

[0177] x6 is the dummy variable conversion value of the maximum number of consecutive months of credit card consumption balance (including consumption installments) in the past three months generated by the feature conversion step;

[0178] x7 is the cube root conversion value of the number of months of credit card recurrence and installment use in the past 6 months generated in the feature conversion step.

[0179] 29. The method according to claim 28, wherein:

[0180] After calculating the credit default probability, the step of calculating the credit score of the sample to be predicted is to use the following formula to calculate and generate a credit score for characterizing the borrower:

[0181]

[0182]

[0183] Among them, P is the borrower's default probability (P) generated in the credit default probability calculation module, A is 54.2458; B is 115.4156, and the round function rounds the calculated score to the integer value; finally, the score greater than 1000 is set to 1000 points, and the score less than 0 is set to 0 points.

[0184] 30. The method according to any one of items 1 to 9, wherein

[0185] The retail credit forecast data is selected from: one, two, three or four of the following: the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, the maximum salary in the past 12 months, the maximum balance of the investment and financial management account in the past 6 months, and the minimum balance of the deposit account at a certain point in time in the past 6 months.

[0186] 31. The method according to item 30, wherein the step of calculating the probability of credit default comprises:

[0187] The average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, the maximum salary in the past 12 months, the maximum balance of the investment and financial management account in the past 6 months, and the minimum balance of the deposit account at a certain point in time in the past 6 months are converted into features.

[0188] Preferably,

[0189] The average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months are converted in a continuous manner; the maximum salary in the past 12 months are converted in a continuous manner; the maximum balance of the investment and financial management account in the past 6 months is converted in a dummy variable manner; the minimum balance of the deposit account at a certain point in time in the past 6 months is converted in a continuous manner;

[0190] Further pre-selected,

[0191] The average monthly asset size of the past 6 months / the average monthly asset size of the past 12 months is converted into the average monthly asset size of the past 6 months / the average monthly asset size of the past 12 months using the original value calculation method; the maximum salary of the past 12 months is converted into the maximum salary of the past 12 months using the natural logarithm calculation method; the minimum balance of the deposit account at a certain point in time in the past 6 months is converted into the minimum balance of the deposit account at a certain point in time in the past 6 months using the natural logarithm calculation method.

[0192] 32. The method according to claim 30, wherein:

[0193] The converted values ​​of the four features, namely, the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, the maximum salary in the past 12 months, the maximum balance of the investment and financial management account in the past 6 months, and the minimum balance of the deposit account at a certain point in time in the past 6 months, are substituted into the sub-model constructed by logistic regression based on the sample retail credit prediction data and credit default probability to calculate the default probability of the sample to be predicted.

[0194] 33. A method according to claim 32, wherein:

[0195] The sub-model is shown in Formula 5 below:

[0196]

[0197] Where k is the number of features entering the model, preferably k is 4;

[0198] α is the intercept term, the value range is (-2.04845, -2.31097), and the optimal value is -2.1797076;

[0199] β1 is the coefficient corresponding to the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, with a value range of (-0.20978, -0.34796) and an optimal value of -0.2788716;

[0200] β2 is the coefficient corresponding to the maximum salary in the past 12 months, with a value range of (-0.05551, -0.39342) and an optimal value of -0.2244648;

[0201] β3 is the coefficient corresponding to the maximum balance of the investment and financial management account in the past 6 months, with a value range of (-0.1207, -0.12475) and an optimal value of -0.122724;

[0202] β4 is the coefficient corresponding to the minimum balance of the deposit account at a certain point in time in the past 6 months. The value range is (-0.68315, -0.93368), and the optimal value is -0.808416.

[0203] x1 is the original value of the average monthly asset size of the past 6 months / the average monthly asset size of the past 12 months generated by the feature conversion step;

[0204] x2 is the natural logarithm-transformed value of the maximum salary in the past 12 months generated in the feature transformation step;

[0205] x3 is the dummy variable conversion value of the maximum balance of the investment and financial management account in the past 6 months generated in the feature conversion step;

[0206] x4 is the natural logarithm-transformed value of the minimum balance of the deposit account at a certain point in time in the past six months generated in the feature conversion step.

[0207] 34. A method according to claim 33, wherein:

[0208] After calculating the credit default probability, the step of calculating the credit score of the sample to be predicted is to use the following formula to calculate and generate a credit score for characterizing the borrower:

[0209]

[0210]

[0211] Among them, P is the borrower's default probability (P) generated in the credit default probability calculation module, A is 54.2458; B is 115.4156, and the round function rounds the calculated score to the integer value; finally, the score greater than 1000 is set to 1000 points, and the score less than 0 is set to 0 points.

[0212] 35. A device for calculating retail credit risk, comprising:

[0213] A data collection module, which is used to obtain retail credit prediction data of samples to be predicted;

[0214] A module for classifying samples to be predicted, which is used to classify the samples to be predicted based on a decision tree method to determine a sub-model for calculating the probability of credit default;

[0215] The credit default probability calculation module is used to substitute the retail credit prediction data into the credit default probability sub-model to calculate the credit default probability of the sample to be predicted.

[0216] 36. An apparatus according to item 35, wherein the apparatus executes the steps of the method for calculating retail credit risk described in any one of items 1 to 34.

[0217] 37. A system for calculating retail credit risk, wherein the system for calculating retail credit risk comprises: a memory, a processor, and a program for the method for calculating retail credit risk stored in the memory and executable on the processor, wherein the program for calculating retail credit risk implements the steps of the method for calculating retail credit risk as described in any one of items 1 to 34 when executed by the processor.

[0218] 38. A computer storage medium, wherein a program for calculating a method for calculating retail credit risk is stored on the computer storage medium, and when the program for calculating a method for calculating retail credit risk is executed by a processor, the steps of the method for calculating retail credit risk as described in any one of items 1 to 34 are implemented.

[0219] 39. A method for constructing a retail credit risk prediction model for credit card business, comprising:

[0220] A data collection step, which obtains the original retail credit prediction data of the sample used to build the model;

[0221] A data derivation step, which processes derived retail credit forecast data based on the original retail credit forecast data;

[0222] The feature initial screening step is to perform initial screening on all categories, that is, all features, including the original retail credit forecast data and the derived retail credit forecast data, to obtain features after initial screening;

[0223] The initial screening data conversion step is to judge the conversion method of the features after the initial screening to confirm whether to use one of the WOE conversion method, the dummy feature conversion method and the continuous conversion method for feature conversion, and use the best method to convert each feature after the initial screening;

[0224] The feature fine screening step is to perform in-depth screening on the initially screened features after feature conversion to obtain finely screened features;

[0225] The credit default probability modeling step is to select the logistic regression method to build the model based on the relationship between the carefully screened features and the probability of credit default, and confirm the method used to calculate the credit default probability;

[0226] The data samples obtained in the data collection step are samples of customers who have used credit card services.

[0227] 40. The method according to claim 39, wherein:

[0228] In the data collection step, the original retail credit prediction data of the samples obtained for building the model include:

[0229] Credit card basic data, which is based on all available data during the sample user's credit card creation and usage process.

[0230] Basic data on personal loans, which is all available data based on the loan application and usage behavior of sample users.

[0231] Basic customer information data, which is based on the attributes of the sample users themselves but is not directly related to their behavior in financial institutions.

[0232] Basic data on personal financial assets, which includes all other financial assets and financial transaction data of sample users in financial institutions that are not related to credit cards and loans.

[0233] 41. The method according to claim 39, wherein:

[0234] In the data derivation step, the derived retail credit prediction data processed based on the original retail credit prediction data refers to data obtained by processing the collected original retail credit prediction data based on the time dimension, space dimension, frequency dimension, and statistical information dimension;

[0235] Preferably, the derived retail credit forecast data includes but is not limited to:

[0236] Derived retail credit forecast data obtained by processing based on sample relationship length,

[0237] Derivative retail credit forecast data obtained by processing time interval variables,

[0238] Derivative retail credit forecast data obtained by processing based on the frequency of sample behavior,

[0239] Derivative retail credit forecast data obtained by processing the current time point of the sample,

[0240] Derivative retail credit forecast data obtained by processing based on sample continuous behavior,

[0241] Derived retail credit forecast data obtained by processing sample data based on statistical information dimensions.

[0242] 42. The method according to any one of items 39 to 41, wherein

[0243] The feature initial screening step includes the following steps:

[0244] The first preliminary screening step is to screen the features based on the missing data of each feature of the sample used to build the model.

[0245] The second initial screening step is to screen the features based on the fact that a single value of a feature sample is too high.

[0246] The third preliminary screening step is to calculate the information IV value of each feature to perform preliminary screening of the features;

[0247] The order of the first preliminary screening step, the second preliminary screening step and the third preliminary screening step can be any order.

[0248] The fourth preliminary screening step uses a stepwise discrimination algorithm to perform preliminary screening of features after the first to third preliminary screenings;

[0249] The fifth preliminary screening step is to conduct preliminary screening of the features after the fourth preliminary screening step based on the consistency between the risk characteristics of each feature itself and the actual real results of the samples used for model construction.

[0250] 43. The method according to any one of items 39 to 42, further comprising:

[0251] The sample selection step is used to screen all users before the data collection step and obtain samples using credit card services for model construction.

[0252] Preferably, the sample selection step includes classifying all sample users based on a decision tree, and the classification basis includes but is not limited to:

[0253] Whether a user is a customer who has applied for credit registration at a financial institution;

[0254] Whether the account age of a certain user is greater than or equal to G1;

[0255] Whether a user is a customer with overdue credit loans;

[0256] Whether the current credit limit usage rate of a certain user is less than a%.

[0257] 44. A method according to any one of items 39 to 43, wherein, in the initial screening data conversion step, the conversion method of the features after the initial screening is determined based on the concentration and data type of the features after the initial screening.

[0258] 45. A method according to claim 44, wherein:

[0259] The initial screening data conversion step includes the following steps based on the judgment of concentration and data type:

[0260] Classify the data type of each feature into character variables and numeric variables.

[0261] For character variables, dummy feature conversion is used to perform initial screening data conversion.

[0262] The process of further classifying numerical variables includes the following sub-steps:

[0263] If the value of the numeric variable is less than n, the WOE conversion method is used to convert the initial screening data.

[0264] If the value of the numerical variable is more than n, further judge if the value of the continuous variable is large and the concentration of a single value is greater than m%, then the WOE conversion method is used; if the concentration of a single value is less than or equal to m%, then the continuous conversion method is used.

[0265] Preferably, n and m are both positive integers, wherein n=5-10, and m=90-99.

[0266] 46. ​​The method according to item 45, further comprising:

[0267] For the features that are confirmed to adopt the continuous conversion method, the optimal conversion method is selected based on the correlation between the feature and credit default under different continuous conversion methods to perform the continuous feature conversion of the feature.

[0268] Preferably, the continuous feature conversion is performed by directly selecting the original value, calculating the square of the original data, calculating the square root of the original data, calculating the cube root of the original data, or calculating the natural logarithm of the original data.

[0269] 47. The method according to any one of items 39 to 46, wherein the feature fine screening step comprises:

[0270] The first fine screening step is to screen the features based on the stepwise regression algorithm, the F test and the T test to determine the significance of the features.

[0271] The second fine screening step is to calculate the variance inflation factor based on each feature and eliminate the features with higher variance inflation factors to screen the features.

[0272] The third fine screening step is to analyze the characteristics after the first fine screening step and the second fine screening step based on logistic regression to see whether the characteristic coefficients are consistent with the trend of the prediction results for credit default so as to further perform feature screening.

[0273] 48. A method according to any one of items 39 to 47, wherein the credit default probability modeling step substitutes the features screened in the feature fine screening step into the Sigmoid function to perform logistic regression to calculate the credit default probability model.

[0274] 49. A device for constructing a retail credit risk prediction model for a credit card business, wherein the device comprises:

[0275] A data collection module, which is used to obtain raw retail credit prediction data of samples used to build a model;

[0276] A data derivation module, which is used to process derived retail credit forecast data based on the original retail credit forecast data;

[0277] A feature initial screening module, which is used to perform initial screening on all categories, that is, all features, including the original retail credit forecast data and the derived retail credit forecast data, to obtain features after initial screening;

[0278] The initial screening data conversion module is used to determine the conversion method of the features after the initial screening to confirm whether to use one of the WOE conversion method, the dummy feature conversion method and the continuous conversion method for feature conversion, and to use the best method for feature conversion for each feature after the initial screening;

[0279] A feature fine screening module is used to perform in-depth screening on the initially screened features after feature conversion to obtain finely screened features;

[0280] The credit default probability modeling module is used to select the logistic regression method to build a model based on the probability relationship between the finely screened features and credit default, and confirm the method used to calculate the credit default probability.

[0281] The data samples obtained by the data collection module are samples of customers who have used credit card services.

[0282] 50. An apparatus according to item 49, wherein the apparatus executes the steps of the method for constructing a retail credit risk prediction model described in any one of items 39 to 48.

[0283] 51. A system for constructing a retail credit risk prediction model for credit card business, wherein the system comprises: a memory, a processor, and a program for constructing a retail credit risk prediction model method stored in the memory and executable on the processor, wherein the program for constructing a retail credit risk prediction model method, when executed by the processor, implements the steps of constructing a retail credit risk prediction model method as described in any one of items 39 to 48.

[0284] Effects of the Invention

[0285] The method and system for constructing a retail credit risk prediction model in this application combines a large number of samples from a large financial institution when constructing the model, and deeply processes and derives the original data obtained from the samples. It uses advanced statistical analysis methods and explainable machine learning technology to construct a general scoring model for retail credit based on the characteristics of the original data and derived data and the strong financial attribute information contained in the data that is difficult to obtain on the market.

[0286] In addition, this application uses the decision tree method to classify samples in the most reasonable way at the beginning of model construction and builds a sub-classification model based on the classified samples. Combining decision trees to build financial models can effectively classify samples according to executable categories, thereby better covering the credit risk characteristics of different customer groups and avoiding the problem of using all samples to build models, resulting in the model lacking sub-group representativeness.

[0287] Furthermore, when constructing the risk prediction model in this application, the overall efficiency of model development is fully improved by using the stepwise discriminant analysis method for the preliminary screening of model features. The stepwise discriminant method can better select more important feature variables under the same dimension, greatly reducing the workload of developers in the next step who need to judge and screen one by one according to the trend of the variables, and fully improving the efficiency of model development without affecting the overall model effect, making the initial screening of variables more efficient and accurate.

[0288] This application innovatively combines continuous conversion, WOE conversion, and dummy feature conversion when building a model, reprocesses some features that have passed the initial screening, combines the advantages and disadvantages of the three conversion methods, and creatively designs a conversion judgment method. According to parameters such as feature data attributes, missing rate, concentration, etc., supplemented by business logic judgment, the optimal feature conversion method is selected. Through such a model building method, it is possible to avoid the technical shortcomings of overfitting when building a model with complete WOE variables and the inability to adapt to categorical variables when building a model with complete continuous variables.

[0289] In addition, the method and system for calculating credit risk probability or scoring credit risk constructed using the model of the present application, namely the Alphad model (Alpha series scoring cards) (including Alphad1 to Alphad5 sub-models or sub-scoring cards), because its model is first split based on the decision tree when it is constructed, so the credit risk is initially calculated by effectively classifying the customer samples and selecting the most suitable sub-model or sub-scoring card for processing. At the same time, because the model construction method used in the sub-model or sub-scoring card is the method described in the present application, the construction process also avoids the technical shortcomings of overfitting when constructing a model with complete WOE variables and the inability to adapt to categorical variables when constructing a model with complete continuous variables. Therefore, it has a significantly better effect than existing models in predicting credit risk.

[0290] This application is applicable to credit business, focusing on the credit card business scenario. It is based on the analysis and design of 130 million customers who have applied for credit business registration, especially using the customer sample of those who applied for credit card business registration for model development, so as to more specifically build a retail risk prediction model for credit card business that is more suitable for commercial banks and other financial institutions. In view of the shortcomings of the existing scoring model, such as the small number of modeling samples, the relatively single data source, the high homogeneity of data dimensions, and the use of mostly weak financial attribute data, which leads to large deviations in prediction results, the model of this application is developed based on highly stable and high coverage data samples. In view of the actual characteristics of the credit card business, more significantly related characteristic variables and more reasonable numerical conversion methods are used to construct the model, so that commercial banking institutions, especially commercial banks with more credit card business, can strengthen credit management and risk control more scientifically and effectively, while also reducing the economic cost of personal credit assessment.

[0291] In this application, different expected levels of customer risk performance are different, and the selection of variables will be very different. For example, customers with an account age of less than 3 have less credit history information, which can be used to simulate new applicants (i.e., those who have not applied for credit business). For example, for the special case of less credit history information for customers with an account age of less than 3, specific parameters are selected to build the model.

[0292] In this application, the credit history information of customers with an account age of less than 6 is aimed at the characteristics of the pre-loan stage of the credit business. According to the different customer information accumulation and risk characteristics, the characteristic variables suitable for the system of this application are selected during modeling. The credit history information of customers with an account age of greater than or equal to 6 and not overdue is aimed at the characteristics of the mid-loan stage of the credit business. According to the different customer information accumulation and risk characteristics, the characteristic variables suitable for the system of this application are selected during modeling. The credit history information of customers with an account age of greater than or equal to 6 and overdue is aimed at the characteristics of the post-loan stage of the credit business. According to the different customer information accumulation and risk characteristics, the characteristic variables suitable for the system of this application are selected during modeling. 90 days overdue is a serious overdue. The probability of a customer repaying the money after the overdue period reaches more than 90 days is low, and the possibility of causing losses to the bank is relatively high. Therefore, this application predicts that the credit overdue is more than 90 days, which can help financial institutions identify high-risk customers, take corresponding measures to prevent them in time, and reduce future losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0293] Figure 1 The graphical results are used to illustrate the model differentiation effect of Example 1 of the present application.

[0294] Figure 2 A and Figure 2 B shows a typical flow chart for classifying all sample users based on a decision tree: Figure 2 A is a decision tree for loan customers; Figure 2 B is used to simulate no loan.

[0295] Figure 3 An example of binning to confirm whether the data conforms to business trends is shown. DETAILED DESCRIPTION

[0296] Credit risk is the risk caused by the borrower's reduced willingness to repay or inability to perform the loan contract due to changes in economic ability, rather than the risk of default caused by the borrower's deliberate fraud. Credit default occurs in all kinds of retail credit business scenarios, and is particularly related to the borrower's personal economic situation. The reasons for borrowers' credit defaults can be divided into four categories: 1. Short credit history, such borrowers have little experience in financial management; 2. Borrowers temporarily forget to repay; 3. Over-borrowing, such borrowers have relatively low repayment ability due to a large amount of debt; 4. Affected by major negative factors, such borrowers have long-term impacts on their repayment ability due to major factors such as reduced income, unemployment or divorce. The above different reasons will more or less cause borrowers to default, which may lead to more serious defaults. The purpose of credit risk scoring is to dig out the inherent mathematical relationship between various historical information of customers and the probability of future default, and convert it into a scoring method to quantify the probability of default.

[0297] At present, the credit risk scoring in the existing technology is mainly developed using credit history data and multi-head data (multi-head data refers to the statistical data of borrowers' behavior of applying for loan demands from multiple financial institutions. It is generally believed that the greater the number of multi-heads in a short period of time, the greater the probability of future overdue payments), which reflects their payment behavior and willingness to pay. The data of the samples used to build the model in this application not only includes transaction behavior information of credit business, but also adds asset data that is usually difficult to obtain. While reflecting the borrower's payment behavior and willingness to pay, it also conducts a more comprehensive measurement of their debt repayment ability, personal qualifications, etc., thereby providing more accurate prediction results.

[0298] <Overall description of model building method>

[0299] Specifically, the present application relates to a method for constructing a retail credit risk prediction model, which includes: a data collection step, which obtains the original retail credit prediction data of the sample used to construct the model; a data derivation step, which processes the derived retail credit prediction data based on the original retail credit prediction data; a feature initial screening step, which performs preliminary screening on all categories including the original retail credit prediction data and the derived retail credit prediction data, that is, all features, to obtain the features after the preliminary screening; a preliminary screening data conversion step, which determines the conversion method of the features after the preliminary screening to confirm that one of the WOE conversion method, the dummy feature conversion method and the continuous conversion method is used for feature conversion, and the best method is used for feature conversion for each feature after the preliminary screening; a feature fine screening step, which performs in-depth screening on the features after the preliminary screening after feature conversion to obtain the finely screened features; a credit default probability modeling step, which selects the logistic regression method for model construction based on the finely screened features combined with the probability relationship between the features and the credit default, and confirms the method for calculating the credit default probability.

[0300] In a specific manner, the method for building a retail credit risk prediction model involved in the present application may further include a sample selection step before the data collection step, which is used to screen all users before the data collection step to obtain samples for model building. Those skilled in the art understand that based on the total number and data of users who can build the model, those skilled in the art can choose whether to remove user data that does not meet the requirements for building the model. In addition, those skilled in the art may also first select some users as samples, and then continuously increase the samples for modeling according to corresponding rules.

[0301] In a specific way, the credit card application and behavior data, personal loan application and behavior data, personal financial asset transaction data, and personal customer information data of retail customers of a financial institution over a period of time can be collected. Some samples can then be eliminated based on the collected sample data. Specifically, for example, the method of this application first collected the credit card application and behavior data, personal loan application and behavior data, personal financial asset transaction data, and personal customer information data of all retail customers of China Construction Bank between 2017 and 2021, totaling 730 million people. The modeling sample is further confirmed through a professional model design scheme, and 630 million data in 2019 are selected as the analysis sample.

[0302] <Sample selection steps>

[0303] In a specific implementation, the sample selection step is to screen all users before the data collection step to obtain samples for model construction. Specifically, the sample selection step includes classifying all sample users based on a decision tree, and the classification basis includes but is not limited to: whether a user is a customer who has applied for credit business registration at a financial institution; whether a user has an account age greater than or equal to G1; whether a user is overdue for credit; whether a user's current repayment rate is less than a%.

[0304] Furthermore, all sample users are classified based on the decision tree. Since performance variables are required for modeling, the analysis samples are further divided into whether they have applied for credit business registration at the relevant bank, that is, users can be divided into those who have applied for registration and those who have not.

[0305] Credit business is a basic and important asset business of banks. It recovers the principal and interest by issuing bank loans and obtains profits after deducting costs. Generally speaking, bank credit business is an important means for banks to make profits. From the classification of bank credit business, it can be divided into corporate credit business and personal credit business. Among them, corporate credit business includes project loans, working capital loans, small business loans, real estate enterprise loans, etc.; personal credit business includes personal housing loans, personal consumption loans, personal business loans, etc.

[0306] Credit card business is the most important credit business of commercial banks. It can create huge profits and revenues for banks. It is also the most convenient and frequently used credit product by retail customers.

[0307] A decision tree is a decision analysis method that uses a decision tree to obtain the probability that the expected value of the net present value is greater than or equal to zero, evaluate project risks, and judge its feasibility based on the known probabilities of various situations. It is a graphical method that intuitively uses probability analysis. When constructing a model, this application uses a decision tree to classify samples used for modeling based on the characteristics of data from large financial institutions. After distinguishing between samples that have applied for credit business and samples that have not applied for credit business, it can further classify whether the samples that have applied for credit business are new customers, and classify the samples that have applied for credit business into two categories: new customers and non-new customers. The decision on whether a customer is a new customer can be determined based on the needs of technical personnel in this field. For example, users who have applied for credit business within m months are decided to be new customers, and customers who have applied for credit business for more than m months are non-new customers.

[0308] In a specific implementation, m may be set to, for example, 3, 6, 9 or 12. In the present application, m may also be referred to as account age.

[0309] Those skilled in the art can fully understand that other splitting methods can also be selected when performing sample splitting. For example, all samples can be first split into customer groups with credit cards and customer groups without credit cards. The customer sample splitting method when building the model can be completely considered and carried out according to the needs of modeling.

[0310] In this application, a sample of credit card customers is first selected for modeling. Using such a sample group can effectively identify the characteristics of such samples, explore the credit and asset characteristics of credit card customers, and build a credit scoring method that is more suitable for commercial banks with credit cards as their main business type. The model built using such a sample can better provide a prediction basis for commercial banks.

[0311] <Original Retail Credit Forecast Data>

[0312] When constructing the model for this application, firstly, based on the basic data of financial institutions, about 120 types of original retail credit prediction data (i.e. more than 120 basic features) can be obtained, and the retail credit risk points are split or processed as much as possible according to different dimensions under the premise of optimal effect, generating more than 3,000 derivative features in total.

[0313] When constructing the model, first of all, based on all historical data of large financial institutions, the credit card application and behavior data, personal loan application and behavior data, personal financial asset transaction data and personal customer information data of all retail customers over the past four years were initially collected when constructing the model of this application, totaling 730 million people, of which each person has corresponding data to be processed every month. It can be seen that the data system used to construct the model of this application is comprehensive and the data volume is very large. When constructing a model based on such a data system, it is necessary to consider the modeling methodology, otherwise it will be trapped in the huge data, resulting in the special sample groups that need to be paid attention to being covered in the huge data volume and unable to be effectively identified, causing the computer program to run slowly or even unable to run, so that the most suitable prediction model cannot be accurately constructed.

[0314] In the data collection step, the original retail credit prediction data of the sample obtained for building the model include: basic data of credit cards, which are all available data based on the sample's credit card creation process and use process (i.e., retail customers' credit card application and behavior data); basic data of personal loans, which are all available data based on the sample's loan application and usage behavior (i.e., personal loan application and behavior data); basic data of basic customer information, which are data based on the attributes of the sample itself but not directly related to the behavior in the financial institution (i.e., personal customer information data); or basic data of personal financial assets, which are all other financial assets and financial transaction data of the sample in the financial institution that are not related to credit cards and loans (personal financial asset transaction data).

[0315] In a specific implementation, the basic data of credit cards include, but are not limited to: credit card bills, credit card cash withdrawals, credit card installments, interest generated by credit cards, and other information such as accounts, overdue, balances, credit limits, due payments, and actual payments under different dimensions (e.g., time dimension, space dimension, and frequency dimension). The above basic data is not limited to the specific categories listed. With the changes in the credit card business, those skilled in the art can further cover new data types in implementation, that is, all types of data that can be obtained during the user's credit card creation and use process can be used as basic data of credit cards.

[0316] In a specific implementation, basic data of personal loans include but are not limited to: personal loan accounts, overdue personal loans, balances of personal loans, total amount of personal loans, repayments due and actual repayments of personal loans, etc. The above basic data are not limited to the specific categories listed. With the changes in personal loan business, those skilled in the art can further cover new data types in implementation, that is, all types of data that can be obtained based on the user's loan application and behavior can be used as basic data of personal loans.

[0317] In a specific implementation, according to the requirements of the Personal Privacy Protection Act, the basic data of the basic information category of customers only includes: basic information of customers, including gender, age and administrative region information of the business. The above basic data is not limited to the specific categories listed. With the development of customer conditions and social relations, technical personnel in this field can also further cover other or emerging data types in the implementation within the scope of the business application scenario, that is, all types of data based on the attributes of the user sample itself but not directly related to the behavior in the financial institution can be used as basic data of the basic information category of customers.

[0318] In a specific implementation, basic data of personal financial assets include but are not limited to: AUM (i.e., asset management scale), deposits, financial management, and payroll information. The above basic data are not limited to the specific categories listed. With the changes in financial assets, those skilled in the art can also further cover new data types in implementation, that is, all other financial assets and financial transaction data that are not related to credit cards and loans in financial institutions based on samples can be used as basic data of personal financial assets.

[0319] In a specific embodiment, the original retail credit prediction data of the samples obtained for building the model are based on more than 120 data types (i.e., basic variables or basic features) obtained from 730 million people, including but not limited to: credit card bills, credit card withdrawals, credit card installments, interest generated by credit cards, and other accounts, overdue, balances, amounts, due and actual repayments under different dimensions (e.g., time dimension, space dimension, frequency dimension); personal loan accounts, overdue personal loan status, personal loan balances, total amount of personal loans, due and actual repayments of personal loans; basic customer information including gender, age and administrative region to which the business belongs; basic information such as AUM (i.e., asset management scale), deposits, wealth management and payroll payment.

[0320] <Derived Retail Credit Forecast Data>

[0321] In the present application, in the data derivation step, the derived retail credit prediction data processed based on the original retail credit prediction data refers to the data obtained by processing the collected original retail credit prediction data based on the time dimension, space dimension, frequency dimension, and statistical information dimension.

[0322] In a specific embodiment, the derived retail credit prediction data includes, but is not limited to: derived retail credit prediction data obtained by processing based on the sample relationship length, derived retail credit prediction data obtained by processing based on time interval variables, derived retail credit prediction data obtained by processing based on the frequency of sample behavior, derived retail credit prediction data obtained by processing based on the current time point of the sample, derived retail credit prediction data obtained by processing based on the continuous behavior of the sample, or derived retail credit prediction data obtained by processing sample data based on statistical information dimensions. For example, the monthly data of the customer can be obtained and processed based on the monthly data. In this application, processing based on statistical information dimensions includes obtaining the maximum value, minimum value, average value, etc. of the data to describe the data situation.

[0323] In a specific implementation, for example, starting from the time dimension, such as customer relationship length variables: such as the customer account opening time, customer maximum account age, etc. are used as types of derived retail credit prediction data, that is, as derived features or derived variables.

[0324] In a specific implementation, for example, starting from the time dimension, time intervals are considered, such as the number of months from the customer's most recent repayment to the current time point, the number of months from the customer's most recent overdue payment to the current time point, etc. as derived features or derived variables.

[0325] In a specific implementation, for example, starting from the frequency level, the behavior frequency level variable is considered: such as the number of times the customer has repaid > N in the last X months, the number of times the customer's credit limit utilization rate > N in the last X months, etc. as derived features or derived variables. There is no limit on X and N, and they can be any positive integer greater than 0 as long as the business logic is reasonable.

[0326] In a specific implementation, for example, starting from the time dimension, the current point in time variables are considered: the customer's current monthly credit limit, the customer's current monthly balance, etc. as derived features or derived variables.

[0327] In a specific approach, starting from the time dimension, continuous behavior variables are considered as derived features or derived variables, such as the maximum number of consecutive overdue payments of a customer > N in the last X months, the number of consecutive repayment rates of a customer > N in the last X months, etc.

[0328] In a specific method, starting from the dimension of statistical information, statistical variables are considered, such as the maximum number of overdue payments of a customer in the last X months, the average credit limit utilization rate of a customer in the last X months, etc. as derived features or derived variables.

[0329] It is clear to those skilled in the art that the above methods for processing derived variables are merely enumerated and can be selected arbitrarily. For example, 120 types of basic data can be processed into more than 3,000 derived data. In this application, data, data types, data types, variables or features are sometimes mixed, and those skilled in the art can understand them based on common sense in statistics.

[0330] In this application, the derivative data can be derived data obtained by simple processing of the original retail credit forecast data, or it can be derived data obtained by complex processing. The simply processed derivative data, such as the current variables such as the salary of the current month and the balance of the current month, can be used directly after the data is aggregated. The complex processed derivative data needs to be based on the variables of the current class based on time slicing and logical processing, and can generate derivative variables such as the maximum salary in the past 3 months and the minimum balance in the past 12 months.

[0331] In this application, the simple processing derived retail credit prediction data or the complex processing derived retail credit prediction data include but are not limited to: the current monthly credit repayment rate, the current remaining available credit limit of the credit card, the number of months from now when the credit limit utilization rate of the credit card in the past 12 months is >90%, the average number of cash withdrawals from the credit card in the past 3 months, the maximum asset size value in the past 3 months, the number of months from now when the minimum balance of the deposit account at a certain point in time in the past 12 months, the number of months with a credit repayment rate >80% in the past 3 months, the number of credit overdue payments >0 in the past 6 months, the current credit card cash withdrawal balance (including cash installments) / current credit limit percentage, the current monthly credit limit utilization rate of revolving loans, the corresponding coefficient of the number of months with interest >0 for special installments in the past 12 months, the number of consecutive months with the largest month-on-month increase in asset size value in the past 3 months, the percentage of overdue interest on mortgage loans in the past 6 months to all repayment amounts in the past 6 months, the credit limit in the past 3 The following items include: percentage of months with monthly credit limit utilization rate > 70%, minimum credit card repayment rate in the past 6 months, number of months since the credit card interest rate > 0 in the past 3 months, percentage of total cash withdrawal amount of credit card consumption in the current month (including installments) / total cash withdrawal amount of consumption in the previous month (including installments), current deposit balance, average balance of investment and financial management accounts in the past 12 months, percentage of months with the largest consecutive increase in asset size in the past 3 months, current remaining available credit card credit limit, percentage of months with credit card repayment rate < 90% in the past 12 months, number of months with the largest consecutive increase in credit card consumption balance (including consumption installments) in the past 3 months, number of months of credit card revolving and installment use in the past 6 months, average monthly asset size in the past 6 months / average monthly asset size in the past 12 months, maximum salary in the past 12 months, maximum balance of investment and financial management accounts in the past 6 months, minimum balance of deposit accounts in the past 6 months.

[0332] <Preliminary feature screening>

[0333] In the initial screening data conversion step of the present application, the judgment of the conversion method of the features after the initial screening is based on the concentration and data type of the features after the initial screening. The initial screening data conversion step includes the following steps based on the judgment of the concentration and data type: classify each feature into character variables and numerical variables according to the data type of each feature, use the dummy feature conversion method to perform initial screening data conversion on the character variables, and further classify the numerical variables. The process includes the following sub-steps: if the value of the numerical variable is less than n, use the WOE conversion method to perform initial screening data conversion, if the value of the numerical variable is more than n, further judge if the value of the continuous variable is more and the concentration of the single value is greater than m%, then use the WOE conversion method, if the concentration of the single value is less than or equal to m%, then use the continuous conversion method, preferably, n and m are both positive integers, where n = 5 to 10, m = 90 to 99.

[0334] For example, n is 5, 6, 7, 8, 9 or 10, and m is 90, 91, 92, 93, 94, 95, 96, 97, 98 or 99.

[0335] Specifically, in this application, a variety of different feature initial screening methods are used to screen a large number of features, so that the screening can be effectively performed from the largest dimension of features. Existing credit risk scoring often uses missing rate, concentration and information value IV for feature initial screening.

[0336] In a specific method, features are screened based on the data missingness of each feature of the sample used to build the model. When screening the missing rate, it is generally considered to delete variables with a missing rate greater than 90%, 91%, 92%, 93%, 94%, or 95%. For example, features with a data missing rate exceeding 95% can be eliminated, and features with a data missing rate exceeding 90% can also be eliminated.

[0337] In a specific method, features are screened based on the fact that a single value of a feature sample is too high. When screening by concentration, it is generally considered to delete variables whose single values ​​account for more than 99%, 98%, 97%, 96%, or 95%. For example, features whose single values ​​exceed 99% can be eliminated, and features whose single values ​​exceed 95% can also be eliminated.

[0338] In a specific method, the information IV value of each feature is calculated to perform preliminary screening of the features. During IV screening, the IV value can be used to measure the predictive power of the feature. The larger the IV value, the stronger the predictive power of the feature. The calculation method of the IV value of a single feature is as follows:

[0339]

[0340] Where k is the number of groups after this feature is discretized; y i is the number of non-defaulting customers in group i; s is the total number of customers who have not defaulted; n i is the total number of defaulting customers in group i; n s The quantitative index of IV has the following meanings: when the calculated value of IV is less than 0.02, it means that the predictive power of this feature is very weak; when the calculated value of IV is above 0.02 and below 0.1, it means that the predictive power of this feature is weak; when the calculated value of IV is above 0.1 and below 0.3, it means that the predictive power of this feature is good; when the calculated value of IV is above 0.3, it means that the predictive power of this feature is strong.

[0341] Of course, the IV calculated value deletion threshold may also be set to 0.03, 0.04, 0.05, etc.

[0342] In the model building method of the present application, the step of using missing rate, concentration and information value IV for feature initial screening can be performed in any order, for example, the screening can be performed based on the missing rate first, then based on the concentration, and finally based on the information value IV for feature initial screening. It is also possible to first screen based on the concentration, then screen based on the missing rate, and finally based on the information value IV for feature initial screening. It is also possible to first screen based on the missing rate, then screen based on the information value IV, and finally based on the concentration for feature initial screening. It is also possible to first screen based on the information value IV, then screen based on the missing rate, and finally based on the concentration for feature initial screening. It is also possible to first screen based on the information value IV, then screen based on the missing rate, and finally based on the concentration for feature initial screening. It is also possible to first screen based on the information value IV, then screen based on the concentration, and finally based on the missing rate for feature initial screening. It is also possible to first screen based on the concentration, then screen based on the information value IV, and finally based on the missing rate for feature initial screening. Those skilled in the art can make a choice based on the data situation of the sample, so based on these three methods, features with obvious defects in one aspect can be effectively removed, which can effectively reduce the data dimension and improve the effect of model building.

[0343] In a specific method, after the features have been screened using missing rate, concentration and information value IV respectively, a stepwise discriminant algorithm is used to perform preliminary screening of the features, and then preliminary screening of the features is performed based on the consistency between the risk characteristics of each feature itself and the actual real results of the samples used to build the model.

[0344] On this basis, the technical solution of the present application introduces a stepwise discriminant method to make the initial screening of variables more efficient and accurate in order to improve the overall efficiency of model development. In actual data, there may be a close distribution of good and bad samples on a certain variable, and the ability to distinguish between good and bad is weak. There may also be a class of variables, each of which can distinguish good and bad samples well, but if all are included in the model, they will be redundant due to the duplication of data dimensions covered by the variables. In this regard, a stepwise discriminant analysis method is used in this application, and WILKS'S LAMBDA value is used as the statistical criterion for entry and removal, and variables with weak or redundant discriminant effects in the data are deleted.

[0345] The stepwise discrimination method uses the WILKS'S LAMBDA criterion to measure the strength of feature capabilities and eliminates features that do not meet the set threshold among the remaining features after three rounds of screening. In the stepwise discrimination process, the variable with the strongest discriminant ability is added first. As the variables in the model gradually increase, the discriminant ability of the variables introduced earlier may also change. If the discriminant ability of a variable in the model is less than the threshold, the variable is removed, and then the process is repeated until the variables contained in the model meet the WILKS'S LAMBDA similarity ratio criterion and other variables do not meet the criteria for entering the model.

[0346] In this application, the step of screening using the stepwise discriminant method is a more critical step. For the purpose of capturing as many credit risk points as possible, the model involved in this application uses a huge amount of data and many data dimensions, so variable screening must be performed to further reduce the time cost of model development. In the existing financial risk model evaluation, the variable importance method is generally used to reduce variables, such as calculating the importance of variables through algorithms such as the Gini index and information entropy, and selecting variables with high importance. The modeling method of screening using the stepwise discriminant method is rarely used. Compared with the variable importance screening scheme commonly used in the industry, the methodology used in this application can retain a large number of variables with relatively weak importance but relatively independent information dimensions.

[0347] In a specific way, more than 3,000 derivative data can be processed based on, for example, 120 types of basic data. After three rounds of screening of missing rate, concentration and information value IV, about 20-30% of the features with poor data can be deleted, but more than 2,000 features will still be retained. If subsequent variable fine screening is performed based on all the above features, it will seriously affect the development efficiency. Therefore, after comparing and testing a variety of different variable deletion schemes, it was finally determined that stepwise discrimination was the best choice.

[0348] For the important features selected by the stepwise discrimination algorithm, further feature selection is performed based on the risk characteristics of each risk point itself (system preset) and the actual bad debt rate of the sample. Determine whether the actual bad debt rate distribution of the remaining features is consistent with the business trend, and eliminate the features whose actual bad debt rate distribution does not conform to the business trend. Specifically, (1) Divide the remaining target features into 10-20 boxes according to the quantile points, and calculate the median value of each box and the corresponding bad debt rate. (2) Use the median value of each box and the previous box and the corresponding bad debt rate to calculate the rate of change (slope). (3) Count the number of boxes greater than 0 and non-0 boxes (non-greater than 0 boxes) in the rate of change between two adjacent boxes, and calculate the percentage of the number of boxes greater than 0 in the rate of change to the number of non-0 boxes in the rate of change. (4) Obtain the risk characteristics preset for this feature, and based on the percentage of the number of boxes greater than 0 in the rate of change to the number of non-0 boxes in the rate of change, check whether the actual bad debt rate change trend is approximately consistent with the business logic bad debt rate change trend. The criteria for whether they are approximately consistent are as follows: First, if the business logic trend of this feature should be: as the value of this feature increases, the bad debt rate increases (such as the credit line utilization rate, etc.). In this module, the features whose number of boxes greater than 0 in the above calculated slope accounts for less than 70% of the number of boxes with a slope other than 0 are eliminated, that is, the features whose feature performance does not conform to the business trend are eliminated. Second, if the business logic trend of this feature should be: as the value of this feature increases, the bad debt rate decreases (such as the deposit amount, etc.). In this module, the features whose number of boxes greater than 0 in the above calculated slope accounts for more than 30% of the number of boxes with a slope other than 0 are eliminated, that is, the features whose feature performance does not conform to the business trend are eliminated.

[0349] For example, the following table gives an example of binning. In this example, the feature values ​​are divided into 10 bins. The value ranges of the bins are also summarized in the following table. At the same time, based on the data in the following table, a graph of the bad debt rate of the feature bins versus the median of the feature bins is drawn, as shown in the following figure: Figure 3 As shown. Figure 3 In the example, there are 7 with slopes greater than 0 and 2 with slopes less than 0. Through the above method, the features can be further screened based on the risk characteristics of each risk point itself (system preset) and the actual bad debt rate of the sample. By further screening features using such a binning method, the features that best match the business development trend can be effectively screened out, thereby further obtaining features suitable for modeling.

[0350]

[0351] <Conversion of eigenvalues>

[0352] The initial feature processing of existing credit risk scoring methods mainly uses WOE conversion (multi-classification) and dummy feature conversion (binary classification) to discretize continuous features (such as age, account age, etc.). The WOE conversion method is an optimal binning scheme based on modeling samples. It discretizes continuous features according to the optimal cutting point and embeds the data of good and bad samples into the WOE value. Therefore, it performs better in model construction, but the binning results may overfit the modeling samples, resulting in a serious decline in model effect when applied to the whole (poor generalization ability). At the same time, due to the normalization operation used in binning, the original features falling into different bins are converted into a single value corresponding to each bin, thus losing the ability to distinguish the risks of people falling into the same interval. The dummy feature conversion method is mainly applied to grouping features. Its advantage is that it can eliminate the difference between good and bad values ​​of different feature values, but it will become extremely complex and redundant when processing continuous variables.

[0353] In the model building method of the present application, the conversion method of the features after preliminary screening is judged to confirm whether to use one of the WOE conversion method, the dummy feature conversion method and the continuous conversion method for feature conversion, and the optimal method is used to perform feature conversion for each feature after preliminary screening.

[0354] The initial screening data conversion step includes the following steps based on the judgment of concentration and data type: classifying each feature into character variables and numeric variables according to the data type of each feature, using dummy feature conversion method to perform initial screening data conversion on character variables, and further classifying the numeric variables includes the following sub-steps: if the value of the numeric variable is less than n, the WOE conversion method is used for initial screening data conversion; if the value of the numeric variable is more than n, further judging that if the value converted to a continuous variable is large and the concentration of a single value is greater than m%, the WOE conversion method is used; if the concentration of a single value is less than or equal to m%, a continuous conversion method is used; preferably, n and m are both positive integers, wherein n=5-10, m=90-99.

[0355] Specifically, taking the character variable as education level, the value of this feature variable can be primary school, middle school, university, graduate school, etc. For numeric variables, if the numeric variable is the number of overdue months in the past three months, the value is 0, 1, 2, 3.

[0356] In the present application, the above n can be 5, 6, 7, 8, 9 or 10, and m can be 90, 91, 92, 93, 94, 95, 96, 97, 98 or 99.

[0357] In a specific implementation, m is selected as 5 and n is selected as 95.

[0358] Specifically, the WOE conversion method is: find the optimal cutting point of the feature, divide the value range of the original feature into multiple bins, calculate the WOE conversion value corresponding to each bin after segmentation according to the good or bad situation in each bin and output it, divide the original feature according to the binning result and output the WOE conversion value. For each bin, the WOE value is calculated as follows:

[0359]

[0360] where y i is the number of non-defaulting customers in group i; s is the total number of customers who have not defaulted; n i is the total number of defaulting customers in group i; n s is the total number of defaulting customers.

[0361] Let's take age as an example to introduce the WOE binning process. The original feature age includes values ​​from 18 to 50 years old. After binning, 5 bins are obtained, namely 18 to 24 years old, 25 to 30 years old, 31 to 35 years old, 36 to 42 years old, and 43 to 50 years old. Then, the WOE conversion value of each bin is calculated according to the number of good and bad customers in each bin. Finally, each user data in each bin is output according to the corresponding conversion value of each bin. For example, 23 years old outputs the WOE conversion value corresponding to the 18 to 24 bins, and 46 years old outputs the WOE conversion value corresponding to the 43 to 50 bins.

[0362] The conversion method of dummy features is: convert a single classification feature into an equal number of dummy features according to the number of values ​​it contains. If a customer belongs to the corresponding value of the generated dummy feature, the corresponding dummy feature value is 1, and the remaining dummy feature values ​​are 0.

[0363] Taking gender as an example, we will introduce the method of dummy feature conversion. The original features include: male, female. After dummy feature conversion, two dummy features, 'gender-male' and 'gender-female', are generated. If the customer's gender is male, 'gender-male' is recorded as 1, and 'gender-female' is recorded as 0. If we take education as an example, the original features include: junior college and below, undergraduate, master and above. After dummy feature conversion, three dummy features, 'education-junior college and below', 'education-bachelor', and 'education-master and above', are generated. If the customer's education is undergraduate, 'education-junior college and below' is recorded as 0, 'education-bachelor' is recorded as 1, and 'education-master and above' is recorded as 0.

[0364] The continuous conversion method is: perform a variety of continuous conversions on the original features (continuous conversion methods include but are not limited to: directly selecting the original value, calculating the square of the original data, calculating the square root of the original data, calculating the cube root of the original data, and calculating the natural logarithm of the original data). Calculate the correlation coefficient r (Correlation Coefficient) between the feature value and the overdue label after continuous conversion, select the conversion method with the largest absolute value of the correlation coefficient, and output the conversion method corresponding to the original feature. The calculation formula of the correlation coefficient is as follows:

[0365]

[0366] Where Σ is the summation symbol in mathematics; n is the total number of observations; x i is the conversion value of the original feature of the i-th observation after continuous conversion; This is the average of the transformed values; where y i is a binary feature indicating whether the i-th observation is in default; It is the average value of this binary feature. The closer the absolute value of the correlation coefficient is to 1, the more relevant the converted value is to the default situation, and the better the effect of this conversion method. The quantitative index of the correlation coefficient r has the following meanings: when the absolute value of the correlation coefficient is above 0 and less than 0.3, it indicates a low correlation; when the absolute value of the correlation coefficient is above 0.3 and less than 0.8, it indicates a moderate correlation; when the absolute value of the correlation coefficient is above 0.8 and less than 1, it indicates a high correlation.

[0367] In the model construction of the present application, for the features that are confirmed to adopt a continuous conversion method, the optimal conversion method is selected to perform continuous feature conversion of the feature based on the correlation of the feature with credit default under different continuous conversion methods. Preferably, the continuous feature conversion is performed by directly selecting the original value, calculating the square of the original data, calculating the square root of the original data, calculating the cube root of the original data, or calculating the natural logarithm of the original data.

[0368] Take age as an example to introduce the continuous conversion method. The original feature age contains values ​​from 18 to 50 years old. The continuous conversion obtains the original value, square, square root, cube root, natural logarithm, etc. of age, and then calculates the WOE value and the absolute value of the correlation coefficient between each conversion value and the good or bad label (prediction result). The conversion method with the largest absolute value of the correlation coefficient is selected for conversion and output. If the cube root of the age feature has the largest absolute value of the correlation coefficient with the good or bad label compared with other conversion methods, the cube root of age is output as the conversion value.

[0369] In the prior art solutions, one of the feature conversion methods, WOE and dummy features, is mainly selected for model construction. For example, in the Chinese patent CN112686749B, the WOE method is used to convert the feature values. The WOE conversion method is an optimal binning scheme based on the modeling sample. It discretizes the continuous features according to the optimal cutting point, so it performs better in model construction, but the binning results may overfit the modeling samples, resulting in a serious decrease in the model effect when applied to the whole (poor generalization ability). At the same time, due to the normalization operation used in the binning, the original features falling into different bins are converted into a single value corresponding to each bin, so the risk differentiation ability of people falling into the same interval is lost.

[0370] The conversion method of dummy features is mainly used for grouping features. Its advantage is that it can eliminate the difference between good and bad values ​​of different feature values. For example, for the industry in which the customer is located, because there is no obvious advantage or disadvantage between the retail industry and the wholesale industry, it is more suitable to use the dummy feature processing method; there is a hierarchical difference between junior colleges and undergraduates, so although dummy features can be used, it is actually more appropriate to use the WOE processing method.

[0371] On the other hand, the continuous conversion method can avoid the overfitting of the modeling samples in the WOE conversion method, has a strong generalization ability for the overall sample, and because no interval mapping is performed, most customer groups rarely fall into a single value. However, it cannot be applied to some features with poor monotonicity or discrete characteristics (such as occupation, position, etc.).

[0372] As described above, the technical solution of the present application takes a different approach and creatively combines three methods: continuous conversion, WOE conversion, and dumb feature conversion. It reprocesses some carefully screened features, combines the advantages and disadvantages of the three conversion methods, and creatively designs a conversion judgment method. It selects the optimal feature conversion method based on parameters such as feature data attributes, missing rate, concentration, and business logic judgment.

[0373] <Logistic regression and deep feature screening based on logistic regression>

[0374] In existing credit risk scoring, due to the requirement of model interpretability, logistic regression model is mainly used for model development, and the software that can be used are generally: SAS, R, Python, etc.

[0375] In a specific embodiment, the present application develops models based on SAS software.

[0376] Specifically, the Sigmoid function is used in logistic regression to fit the probability of predicted default. The Sigmoid function is:

[0377]

[0378] Where Z is a linear combination of the model coefficients and the feature transformation values, and Z is defined as follows:

[0379] Z=α+β1x1+β2x2+...+β k-1 x k-1 +β k x k

[0380] The predicted probability of default is:

[0381] P=P(Y=1|x1,x2,x3,...,x k-1 , x k )

[0382] The fitted prediction is that the probability of default is:

[0383]

[0384] From the above formula we can further deduce:

[0385] Substituting the Z value into the above formula can calculate the probability P of predicted default.

[0386] The core of logistic regression model construction is feature screening. The steps of feature screening are as follows: First, batch screen the features according to missing rate, concentration and information value IV. Second, screen all remaining features one by one according to whether the features are in line with business trends, and retain the features with correct business trends. For example: if it is found that the bad debt rate of the customer group decreases with the increase of the characteristic loan balance, it is considered that this feature does not conform to the business trend. In the understanding of credit business, the higher the loan balance, the higher the customer's default risk exposure (EAD) level, and the greater the risk. At this time, this feature will be removed from the feature list. Third, use the stepwise regression function of logistic regression to eliminate the less important features that are highly correlated with other features. Fourth, filter the coefficients according to the positive and negative signs of the training coefficients and the business trend of the feature conversion value, and retain the features whose feature coefficient signs conform to the business logic. In the case where 0 is defined as a good customer and 1 is defined as a bad customer in the Y label, for a feature whose bad debt rate increases monotonically with the increase of the feature value (such as loan balance), the sign of its training coefficient should be positive; conversely, for a feature whose bad debt rate decreases monotonically with the increase of the feature value (such as deposit balance), the sign of its training coefficient should be negative. If it does not meet the above standards, it should be eliminated. Fifth, use variance inflation factor (VIF), correlation coefficient, etc. to further eliminate features with high correlation: for variance inflation factor, eliminate features with the highest VIF and greater than 4 one by one; for features with high correlation, eliminate features with low IV values ​​in the feature group with the highest correlation coefficient and greater than 0.80 one by one. Sixth, use the population stability index (PSI) to eliminate features with large distribution differences at different time points, which lead to instability. For features with PSI>0.25, directly eliminate them, and for features with 0.25>PSI>0.1, carefully eliminate them according to the impact of eliminating this feature on the model's distinguishing ability.

[0387] This application strictly adheres to the above rules to screen features, ensuring the model's interpretability, stability, and ability to distinguish between good and bad customers.

[0388] The feature fine-screening steps of the present application include: a first fine-screening step, based on a stepwise regression algorithm, to screen features based on the F test and the T test for the significance of features; a second fine-screening step, to screen features based on calculating the variance inflation factor for each feature and eliminating features with higher variance inflation factors; a third fine-screening step, based on logistic regression, to analyze whether the feature coefficients of the features after the first fine-screening step and the second fine-screening step are in line with the trend of the prediction results for credit default so as to further perform feature screening.

[0389] Feature fine screening step 1, based on the stepwise regression algorithm, based on the F test and T test method, introduces features in order from high to low significance. Each time a feature is introduced, the selected features are tested one by one. When the originally introduced feature becomes no longer significant due to the introduction of the subsequent features, it is eliminated. This process is repeated until no features above the significance threshold are selected into the equation, and no features below the significance threshold are eliminated from the regression equation.

[0390] Feature fine-screening step 2 is based on the method of eliminating features with high variance inflation factors to further reduce multicollinearity in the model.

[0391] Feature fine screening step 3, based on the risk characteristics of each risk point itself (system preset) and the positive and negative signs of the model training coefficient, determines whether the feature coefficients of the remaining features in the model in the feature fine screening step 3 are consistent with the business trend, removes the features whose model coefficients do not meet the business trend, and iterates again. The specific implementation plan of feature fine screening module 3 is as follows: 1. For features whose feature conversion method is WOE type, the corresponding model training coefficient should be a negative value, and the WOE conversion type features with positive training coefficients should be eliminated. 2. For continuous conversion methods, if the value of this feature increases in business logic, the bad debt rate should increase (such as the credit line utilization rate, etc.), then the corresponding model training coefficient should be a positive value, and such continuous conversion type features with negative training coefficients should be eliminated; if the value of this feature increases in business logic, the bad debt rate should decrease (such as the deposit amount, etc.), then the corresponding model training coefficient should be a negative value, and such continuous conversion type features with positive training coefficients should be eliminated. 3. For the dummy feature conversion method, it is necessary to determine the bad debt rate when the value is 1 and the value is 0. If the bad debt rate when the value is 1 is greater than the bad debt rate when the value is 0, the coefficient should be positive, otherwise it should be negative.

[0392] Feature fine screening step 4, data stability monitoring step, is used to evaluate whether the distribution of individual features and overall scores at different time points has obvious deviations. In this embodiment, features with PSI>0.25 are directly eliminated, and features with 0.25>PSI>0.1 are carefully eliminated based on the impact of eliminating this feature on the model's distinguishing ability.

[0393] Based on the above steps 3 and 4, the feature list is iterated multiple times until no new features are added or removed from the model, and the iteration is stopped to obtain the final feature list and its conversion value. After the above steps, the input variables for constructing the model of this application can be finally obtained.

[0394] The credit default probability modeling step of the present application substitutes the features screened in the feature fine screening step into the Sigmoid function to perform logistic regression to calculate the credit default probability model.

[0395] In the prior art, the core of all scoring models built is whether the data they use is representative. For the existing scoring models in the prior art, due to information security and cost reasons, the sample size and bad labels of most models are small, and the stability of the model cannot be guaranteed, let alone independent modeling of a certain type of customer. At the same time, because the scoring models on the market can obtain information dimensions in the modeling process, mostly multi-loan data and weak financial attribute data (such as smart terminal device data, social platform data, online shopping mall data and other non-credit transaction data), it is impossible to accurately reflect the customer's asset status and repayment ability. The data source used in this application is the full business data of large banks, and the data volume of modeling samples and bad labels is very large. This application designs modeling samples based on different sample groups, which can more finely distinguish the risk differences between such customers, and at the same time ensure the stability of data and models based on cross-time verification and PSI verification. This application uses personal credit history data and asset data for model development, so it has a good reflection of the borrower's repayment willingness and repayment ability.

[0396] For some small and medium-sized financial institutions, the internally built scoring models are relatively small or started late in their retail credit business, and the accumulated historical data is insufficient to develop a credit risk model with stable data and strong differentiation ability. Therefore, they are highly dependent on manual approval when used for credit review business. The efficiency limitations of manual approval have restricted the development of their retail credit business. At the same time, the subjectivity of manual approval has increased the operational risks in the credit review process. The model constructed in this application can assist such financial institutions in making digital decisions, enhance their approval accuracy and speed, and reduce the above-mentioned adverse effects.

[0397] In terms of feature conversion, compared with the traditional WOE conversion method, it is necessary to roughly bin and discretize continuous features according to data performance and the experience of modelers. The results of rough binning are greatly affected by the subjective factors of modelers; and the process of discretizing continuous features may result in too many customers falling into the same interval, which leads to more single values ​​of the calculated scores. This application combines the continuous conversion method, the WOE conversion method and the dummy feature conversion method to encode the original features, which reduces the impact of human factors and single values ​​while enhancing the ability to distinguish scores.

[0398] <Methodology for calculating retail credit risk>

[0399] The present application further relates to a method for calculating the retail credit risk of a sample to be tested using the retail credit risk model constructed in the present application.

[0400] In this application, retail credit risk refers to the probability of credit default in the sample to be tested. In a specific manner, the probability of credit default refers to the probability of credit default, specifically, for example, whether a credit default of more than 90 days will occur.

[0401] The present application relates to a method for calculating retail credit risk, which comprises: a data collection step, which obtains retail credit prediction data of samples to be predicted; a step of classifying the samples to be predicted, which classifies the samples to be predicted based on a decision tree method to determine a sub-model for calculating the probability of credit default; a step of calculating the probability of credit default, which substitutes the retail credit prediction data into the credit default probability sub-model to calculate the probability of credit default of the samples to be predicted.

[0402] In the present application, the sample to be predicted may be a sample used to build a model, that is, when the model is built, the sample is already a customer of a certain financial institution holding a credit business (including but not limited to a credit card or personal loan business), and the method for calculating retail credit risk of the present application can be used to calculate its potential retail credit risk in the future. In the present application, the sample to be predicted may be a sample that is not used to build a model, that is, when the model is built, it is not a customer of a certain financial institution holding a credit business, but now it is a customer of a certain financial institution holding a credit business, and the method for calculating retail credit risk of the present application can be used to calculate its potential retail credit risk in the future. In the present application, the sample may also be a customer who was not and is not a customer of the financial institution holding a credit business when the model was built in the past, but is a customer of the financial institution that already holds other financial asset businesses, and the method for calculating retail credit risk of the present application can be used to calculate its potential retail credit risk in the future.

[0403] The present application relates to a method for calculating retail credit risk, which comprises: a data collection step, which obtains retail credit prediction data of samples to be predicted; a step of classifying the samples to be predicted, which classifies the samples to be predicted based on a decision tree method to determine a sub-model for calculating the credit default probability; a credit default probability calculation step, which substitutes the retail credit prediction data into the credit default probability sub-model to calculate the credit default probability of the samples to be predicted; and after calculating the credit default probability, a step of calculating the credit score of the samples to be predicted.

[0404] The step of calculating the credit score of the sample to be predicted is used to calibrate the calculated credit default probability to a standardized score of 0-1000 points.

[0405] In the method of the present application, the retail credit prediction data includes the original retail credit prediction data of the sample to be predicted and the derivative retail credit prediction data processed based on the original retail credit prediction data. The descriptions of the original retail credit prediction data and the derivative retail credit prediction data are consistent with those in the model construction section.

[0406] In the present application, the steps of classifying the sample to be predicted include the following sub-steps: whether the sample to be tested is a customer who has applied for a credit service at a financial institution; whether the sample to be tested has an account age greater than or equal to G1; whether the credit of the sample to be tested is overdue; whether the current repayment rate of the sample to be tested is less than a%; classifying the sample to be predicted based on the above sub-steps to determine the sub-model for calculating the credit default probability, and the order of the above sub-steps can be set arbitrarily.

[0407] In the present application, Figure 2 A and Figure 2 B shows the flow chart of classifying the sample to be predicted. Based on the classification of this step, the sub-model for calculating the credit default probability can be selected. As Figure 2 A and Figure 2 B shows, the sample to be predicted is classified in the following order: The first case: According to Figure 2 B, it is judged whether the sample to be tested is a customer who has applied for a registered credit service at a financial institution, that is, it is judged whether the account age is less than G2 (G2 is, for example, 1, 2, 3, and is preselected as 3); The second case: According to Figure 2 A, it is judged whether the sample to be predicted is a customer with an account age greater than or equal to G1 (G1 is, for example, 1, 2, 3, 4, 5, 6, 7, 8, 9, and is preferably 6); Then it is judged whether the sample to be tested is a customer with a credit loan overdue; Then it is judged whether the utilization rate of the sample to be tested is < a% (for example, a can be 50, 60, 70, 80, 90, 95, 96, 97, 98, 99, 100, and is preferably 100). There is no order of priority between the first case and the second case, and it can be selected according to actual needs. Using such step classification can confirm the most suitable sub-model for predicting the sample to be tested.

[0408] In the present application, after performing feature transformation on the retail credit prediction data, it is substituted into the credit default probability sub-model to calculate the credit default probability of the sample to be predicted. The feature transformation step includes selecting the WOE method or the continuous type method for feature transformation based on the feature type of the retail credit prediction data to be substituted into the credit default probability sub-model.

[0409] The methods for feature transformation for the WOE method or the continuous type method are well-known to those skilled in the art. The specific transformation methods can refer to the description in the model construction section of the present application. However, the existing models in the art only use the WOE method for parameter transformation.

[0410] The WOE conversion method is an optimal binning scheme based on the modeling sample. It discretizes the continuous features according to the optimal cutting point. Therefore, it performs better in model construction, but the binning results may overfit the modeling sample, resulting in a serious decline in the model effect when applied to the overall population (poor generalization ability). At the same time, due to the normalization operation used in binning, the original features falling into different bins are converted into a single value corresponding to each bin, thus losing the ability to distinguish the risks of people falling into the same interval. The conversion method of dummy features is mainly applied to grouping features. Its advantage is that it can eliminate the difference between good and bad values ​​of features. For example: for the industry in which the customer is located, because there is no obvious advantage or disadvantage between the retail industry and the wholesale industry, it is more suitable for the dummy feature processing method; there is a hierarchical difference between junior colleges and undergraduates, so although dummy features can be used, it is actually more appropriate to use the WOE processing method. The continuous conversion method can avoid the overfitting of the modeling sample in the WOE conversion method, and has a strong generalization ability for the overall sample. Since no interval mapping is performed, most customer groups rarely fall into a single value. However, it cannot be applied to some features with poor monotonicity or discreteness (such as occupations, positions, etc.). The present application uses a combination of continuous conversion and WOE conversion to process the initial features, combines the advantages and disadvantages of the two conversion methods, designs a conversion judgment module, and selects the optimal feature conversion method based on the parameters such as feature data attributes, missing rate, and concentration, supplemented by business logic judgment. The technical solution of the present application takes a different approach, creatively combining the three methods of continuous conversion, WOE conversion, and dummy feature conversion, reprocesses some carefully screened features, combines the advantages and disadvantages of the three conversion methods, and creatively designs a conversion judgment method, and selects the optimal feature conversion method based on the parameters such as feature data attributes, missing rate, and concentration, supplemented by business logic judgment. The feature conversion method determined in this way is then substituted into the model for calculating retail credit risk, which can more accurately predict the retail credit risk of the sample.

[0411] The sub-model used in the present application to calculate the probability of credit default is a model constructed based on sample retail credit prediction data and the probability of credit default using logistic regression based on an existing user population, that is, a model constructed using the method described in the present application.

[0412] <Device, system and computer storage medium for calculating retail credit risk>

[0413] The present application relates to a device for calculating retail credit risk, which includes: a data acquisition module, which is used to obtain retail credit prediction data of samples to be predicted; a module for classifying samples to be predicted, which is used to classify samples to be predicted based on a decision tree method to determine a sub-model for calculating credit default probability; a credit default probability calculation module, which is used to substitute the retail credit prediction data into the credit default probability sub-model to calculate the credit default probability of the samples to be predicted.

[0414] The device for calculating retail credit risk of the present application can execute the steps of the method for calculating retail credit risk of the present application.

[0415] The present application also relates to a system for calculating retail credit risk, wherein the system for calculating retail credit risk comprises: a memory, a processor, and a program for calculating retail credit risk stored in the memory and executable on the processor, wherein the program for calculating retail credit risk implements the steps of the method for calculating retail credit risk as described in the present application when executed by the processor.

[0416] The present application relates to a computer storage medium, wherein a program for a method for calculating retail credit risk is stored on the computer storage medium, and when the program for the method for calculating retail credit risk is executed by a processor, the steps of the method for calculating retail credit risk as described in the present application are implemented.

[0417] All the contents described above for the method for calculating retail credit risk can be fully applied to the device, system and computer storage medium for calculating retail credit risk.

[0418] The method for calculating retail credit risk in the present application can avoid the technical shortcomings of overfitting when building a model with complete WOE variables and the inability to adapt to categorical variables well when building a model with complete continuous variables. Therefore, when the model is used for retail credit risk calculation, it can better cover the retail credit risk prediction needs of various types of customer groups and meet the demand for obtaining more accurate prediction results when predicting credit risk as a general score.

[0419] Example 1 Collection of modeling samples

[0420] In the process of constructing this embodiment, the credit card application and behavior data, personal loan application and behavior data, personal financial asset transaction data and personal customer information data of all retail customers of China Construction Bank between 2017 and 2021 were collected, totaling 730 million people. The modeling sample was confirmed through a professional model design scheme, and 630 million data in 2019 were selected as the analysis sample. Since performance variables are required for modeling, the analysis sample is further divided into two parts: those who have applied for registered credit business and those who have not applied for registered credit business, with sample sizes of 130 million and 500 million people respectively.

[0421] When designing the specific model, analysis and design will be conducted on the 130 million customers who have already applied. After the model is developed and put online, the results will be applied to the entire 730 million population, whether they have already applied or not, or to all other potential customer groups.

[0422] The model design includes 1) exclusion rules: such as excluding customer data that has been settled, closed, has no performance, or has special circumstances. After exclusion, the modeling population is 58.5 million customers (credit card customers); 2) time window setting: determine the sample to use 2019 data as the modeling sample, and use the next 15 months as the performance period of Y; 3) sample sampling: use a good / bad sample number of 5:1 for sampling modeling. Different segmentation schemes are designed through the decision tree method, and the parent-child model comparison method is adopted to confirm the final model segmentation scheme, and then each segmentation model is modeled separately. The model segmentation scheme of this application is designed by dividing the customer's account age by whether it is less than 6, whether it is currently overdue, and whether the current credit limit utilization rate is greater than or equal to 100%. It fully reflects the control variables that are closely related to risk characteristics in the process of business development, and can more comprehensively match various customer groups in the market.

[0423] In this embodiment, the modeling samples are first divided into the following sub-model modeling samples based on the decision tree for the construction of subsequent sub-models. The first layer of the decision tree when determining the modeling samples is overdue, which is used to distinguish the customer's historical repayment behavior. In this embodiment, for customers with an account age greater than or equal to 6, the user population is non-repeatedly divided into three models: Alphad2, Alphad3, and Alphad4 according to their current number of overdue periods and credit line utilization rate; customers with a credit account age less than 6 are divided into Alphad1, where the bad debt rate refers to the ratio of bad customers in a certain type of sample to the total number of samples in this type.

[0424] For customers who are not currently overdue, based on their current credit utilization rate: if the current number of overdue periods is greater than 0, then enter the Alphad2 segmentation model; if the current number of overdue periods is equal to 0 and the current credit utilization rate is greater than or equal to 100%, then enter the Alphad3 segmentation model; if the current number of overdue periods is equal to 0 and the current credit utilization rate is less than 100%, then enter the Alphad4 segmentation model

[0425] In addition, in order to build a model for the sample group that has not applied for credit business registration, in this embodiment, the account age is used as a segmentation variable, and only the customer group with an account age of less than 3 months is selected to enter the model Alphad5.

[0426] Specifically, the sample size of a sub-model used to construct the Alphad 1 segmentation model is about 30,000. Its customer base is mainly those with an account age of less than 6 years. The model is used to predict the probability of credit delinquency of more than 90 days in this group of people.

[0427] Specifically, the sample size of a sub-model used to build the Alphad 2 segmentation model is about 1 million. Its customer base is mainly those with an account age greater than or equal to 6 and are currently overdue. The model is built to predict the probability of this group of people being overdue for credit for more than 90 days.

[0428] Specifically, the sample size of a sub-model used to build the Alphad 3 segmentation model is about 360,000. Its customer base is mainly those with an account age of more than 6 years and a current credit limit utilization rate greater than or equal to 100%. The model is built to predict the probability of credit delinquency of more than 90 days in this group of people.

[0429] Specifically, the sample size of a sub-model used to build the Alphad 4 segmentation model is about 3 million. Its customer base is mainly those with an account age of more than 6 and a current credit limit utilization rate of less than 100%. The model is built to predict the probability of credit delinquency of more than 90 days in this group of people.

[0430] Specifically, the sample size of a sub-model used to build the Alphad 5 segmentation model is about 20,000. Its customer base is mainly those who have applied for credit business and have an account age of less than 3 years. The model is built to predict the probability of this group of people being overdue for credit for more than 90 days.

[0431] Based on the different customer samples of each sub-model confirmed above, historical data information of borrowers is obtained: 1) Credit card category, basic fields include account, overdue, balance, quota, repayment, actual repayment and other information under different dimensions such as bills, cash withdrawals, installments, interest, etc.; 2) Personal loan category, including account, overdue, balance, quota, repayment, actual repayment and other information; 3) Customer basic information, including gender, age and administrative region information of the business; 4) Personal financial assets, including AUM, deposits, wealth management and payroll information.

[0432] According to various risk points in the credit business, macro credit risk is split according to information dimensions, time slices, etc. Variables are derived through professional characteristic variable construction methods. The basic methods are: 1) Customer relationship length variables: such as customer account opening time, customer maximum account age, etc. 2) Time interval variables: such as the number of months from the customer's last repayment to the observation point, the number of months from the customer's last overdue payment, etc.; 3) Behavior frequency degree variables: such as the number of times the customer's repayments are > N in the last X months, the number of times the customer's credit utilization rate is > N in the last X months, etc.; 4) Current time point variables: customer's current monthly credit limit, customer's current monthly balance, etc.; 5) Statistical value variables: such as the customer's maximum number of overdue periods in the last X months, the customer's average credit utilization rate in the last X months; 6) Continuous behavior variables: such as the maximum number of consecutive overdue periods > N in the last X months, the number of consecutive repayment rates > N in the last X months, and other characteristic variables. Finally, 3,067 features with potential predictive power for customer overdue payments were derived, such as "current repayment amount", "average number of overdue periods in the past 6 months", and "number of credit card installments in the past 12 months". In this article, observation point refers to the time point when samples are collected until modeling, and current also refers to the sampling cutoff time. Observation point and current have the same meaning.

[0433] Table 1 Summary of basic variables and derived variables used in the examples

[0434]

[0435]

[0436] Example 2 Feature Screening

[0437] A preliminary screening was performed on the 3067 features (variables) collected in Example 1 that have potential predictive power for customer overdue payments.

[0438] The preliminary screening of the features of the four sub-models Alphad 1 to Alphad 4 divided in Example 1 is performed according to the following method:

[0439] In the first round of preliminary screening, features with a missing rate of more than 95% in the data collected in Example 1 were first eliminated, and a total of 645 variables were deleted, leaving 2422 variables.

[0440] In the second round of preliminary screening, for the 2214 features that passed the first round of preliminary screening, the features with single values ​​exceeding 99% among the remaining features were eliminated, and a total of 207 variables were eliminated, leaving 2007 variables.

[0441] In the third round of preliminary screening, the 2007 features after the second round of preliminary screening were sorted according to the feature values ​​(the sorting method is specifically that if the feature is a character variable, each value is a separate box, and if the feature is a numeric variable, it is sorted from small to large according to the value size), and then divided into 10-20 boxes according to the quantile point, and the feature IV value was calculated. If the feature IV value is lower than 0.02, it will be eliminated. In this embodiment, since the variable attributes are strong financial and asset variables, the prediction effect is strong. Here, variables with IV values ​​lower than 0.05 are eliminated. A total of 873 features are eliminated, and 1134 feature variables remain.

[0442] In the fourth round of preliminary screening, the 1134 features after the third round of screening were further screened based on the stepwise discrimination algorithm. After this round of screening, 303 important feature variables can be quickly screened out. The following is an introduction to the stepwise discrimination method. In actual data, there may be little difference in the mean values ​​of different categories on a certain variable. In this case, the effect of using this variable for classification will not be very good; there is also a type of variable that can indeed distinguish different categories in the data well when considered independently, but after including these variables in the model variables, they may appear redundant. Therefore, the embodiment pioneered the use of the stepwise discrimination method to better select more important feature variables under the same dimension, greatly reducing the workload of judging and screening one by one according to the trend of the variables in the fifth round of screening, and fully improving the efficiency of model development without affecting the overall model effect.

[0443] The stepwise discrimination method uses the WILKS'S LAMBDA criterion to measure the strength of feature capabilities and eliminates features that do not meet the set threshold among the remaining features after three rounds of screening. In the stepwise discrimination process, the variable with the strongest discriminant ability is added first. As the variables in the model gradually increase, the discriminant ability of the variables introduced earlier may also change. If the discriminant ability of a variable in the model is less than the threshold, the variable is removed, and then the process is repeated until the variables contained in the model meet the WILKS'S LAMBDA similarity ratio criterion and other variables do not meet the criteria for entering the model.

[0444] The fifth round of preliminary screening is to further screen the 303 important features after the fourth round of screening based on the risk characteristics of each risk point itself (system preset) and the actual bad debt rate of the sample.

[0445] Determine whether the actual bad debt rate distribution of the remaining features is consistent with the business trend, and remove the features whose actual bad debt rate distribution of the samples does not conform to the business trend. Specifically,

[0446] (1) Divide the remaining target features into 10-20 boxes according to the quantiles, and calculate the median of each box and the corresponding bad debt rate.

[0447] (2) Calculate the rate of change (slope) using the median of the values ​​of each box compared to the previous box and the corresponding bad debt rate.

[0448] (3) Count the number of boxes with a change rate greater than 0 and the number of boxes with a change rate greater than 0 between two adjacent boxes, and calculate the percentage of boxes with a change rate greater than 0 to the number of boxes with a change rate greater than 0.

[0449] (4) Obtain the risk characteristics preset for this feature, and based on the percentage of the number of boxes greater than 0 in the above-calculated change rate to the number of non-zero boxes in the change rate, determine whether the actual bad debt rate change trend is approximately consistent with the business logic bad debt rate change trend.

[0450] The criteria for approximate consistency are as follows:

[0451] First, if the business logic trend of this feature is: as the value of this feature increases, the bad debt rate increases (such as the credit line utilization rate, etc.), in this module, the features whose number of boxes with slope greater than 0 in the above calculation accounts for less than 70% of the number of boxes with slope non-0 are eliminated, that is, the features that do not conform to the business trend are eliminated.

[0452] Second, if the business logic trend of this feature is: as the value of this feature increases, the bad debt rate decreases (for example, the amount of deposits, etc.), in this module, the features whose number of boxes with slope greater than 0 in the above calculation accounts for more than 30% of the number of boxes with slope not 0 are eliminated, that is, the features that do not conform to the business trend are eliminated.

[0453] After the fifth round of screening, 89 features were eliminated, leaving 214 features.

[0454] For the Alphad 5 sub-model divided in Example 1, the selected variables are customer basic information and personal financial asset variables (a total of 316 feature variables) used to approximately predict the probability of credit default in the future for customers who have not registered for credit applications. The feature initial screening step and subsequent steps are the same as other sub-models.

[0455] Example 3 Conversion of features after initial screening

[0456] In the feature judgment step, the optimal conversion method is selected according to the concentration of features after the initial feature screening step, data type, etc. When judging the conversion method, data types are generally divided into two categories, one is character variables, and the other is numeric variables. For character variables, the dummy variable conversion method is generally used for variable conversion; for numeric variables, if the variable value is less than 5 or so, the WOE conversion method will be used. If the variable value is more, the WOE or continuous conversion method will be used (the optimal conversion form is selected by the correlation with the target variable). In this process, it is necessary to comprehensively consider the situation of variable concentration. For example, if the continuous variable has many values, but the concentration of a single value exceeds 95%, the continuous processing will not be performed, and the WOE conversion method can be used directly. Features using different conversion methods are grouped and divided into different data sets. Specifically,

[0457] First, the remaining 214 features in Example 2 are used to determine the conversion methods of these features. The following three methods are selected based on the concentration of the features, data types, etc.

[0458] Feature conversion method 1 is used to perform WOE conversion on the features obtained in the data acquisition module and judged in the feature judgment module that the optimal conversion method is WOE conversion.

[0459] Feature conversion method 2 is used to perform dummy feature conversion on the features obtained in the data acquisition module and judged in the feature judgment module that the optimal conversion method is dummy feature conversion.

[0460] Feature conversion method 3 is used to obtain the features in the data acquisition module that are determined to be optimal for continuous conversion in the feature judgment module, select the optimal continuous conversion method, and perform continuous conversion.

[0461] The feature merging module is used to horizontally splice the data of feature conversion mode 1, feature conversion mode 2 and feature conversion mode 3.

[0462] Through this embodiment, 214 features were converted, of which 24 features were converted to WOE, 45 features were converted to dummy features (expanded to 15 variables), and 145 features were converted to continuous types. The conversion of the features after the initial screening of the Alphad5 sub-model is also basically similar.

[0463] Example 4 Feature Depth Screening (Feature Fine Screening Step)

[0464] The feature fine screening step mainly includes the following four steps in this embodiment. Steps 1 and 2 are based on stepwise regression and calculation of variance inflation factor to eliminate features with large multicollinearity in the feature merging module to enhance the robustness of the model. Step 3 eliminates features whose positive and negative signs of the training coefficients in the model do not conform to the business trend. Step 4 uses the population stability coefficient (PSI) to eliminate unstable features. This embodiment uses the LOGISTIC process in SAS for deep screening.

[0465] Feature fine screening step 1, based on the stepwise regression algorithm, based on the F test and T test method, introduces features in order from high to low significance. Each time a feature is introduced, the selected features are tested one by one. When the originally introduced feature becomes no longer significant due to the introduction of the subsequent features, it is eliminated. This process is repeated until no features above the significance threshold are selected into the equation, and no features below the significance threshold are eliminated from the regression equation. After this step, 214 features are eliminated to 108 features.

[0466] Feature fine-screening step 2, which is based on the method of removing features with high variance inflation factors, further reduces multicollinearity in the model. After this step, 108 features are eliminated to 65 features.

[0467] Feature fine screening step 3 compares the risk characteristics of each risk point itself (system preset) with the positive and negative signs of the model training coefficients to determine whether the feature coefficients of the remaining features in the model in step 3 are consistent with the business trend, eliminates the features whose model coefficients do not conform to the business trend, and iterates again.

[0468] The specific implementation scheme of the feature fine screening module 3 is as follows:

[0469] 1. For features whose feature conversion method is WOE type, the corresponding model training coefficient should be a negative value, and the WOE conversion type features with positive training coefficients should be eliminated.

[0470] 2. For continuous conversion methods, if the value of this feature increases in business logic, the bad debt rate should increase (such as the credit limit utilization rate, etc.), then the corresponding model training coefficient should be positive, and such continuous conversion features with negative training coefficients should be eliminated; if the value of this feature increases in business logic, the bad debt rate should decrease (such as the deposit amount, etc.), then the corresponding model training coefficient should be negative, and such continuous conversion features with positive training coefficients should be eliminated.

[0471] 3. For the dummy feature conversion method, it is necessary to determine the bad debt rate when the value is 1 and the value is 0. If the bad debt rate when the value is 1 is greater than the bad debt rate when the value is 0, the coefficient should be positive, otherwise it should be negative.

[0472] Feature fine screening step 4, data stability monitoring step, is used to evaluate whether the distribution of individual features and overall scores at different time points has obvious deviations. In this embodiment, features with PSI>0.25 are directly eliminated, and features with 0.25>PSI>0.1 are carefully eliminated based on the impact of eliminating this feature on the model's distinguishing ability.

[0473] Based on the above steps 3 and 4, the feature list is iterated multiple times until no new features are added or removed from the model, and the iteration is stopped to obtain the final feature list and its conversion value. After the above steps, 65 features are eliminated as the feature variables finally entered into the model, for example, 5 features, 6 features, 7 features, and 8 features.

[0474] Example 5 Construction of Alphad 1 Model

[0475] Generally speaking, the stronger the correlation between the feature variable and the target variable, the better the accuracy of the final model. In this embodiment, Alphad 1 is a sub-model of the segmentation model, such as Figure 2 As shown in A, by adopting the above decision tree method, the sample size obtained by classification is about 30,000, and its customer base is mainly customers with an account age of less than 6 years. The model is built to predict the probability of credit overdue for more than 90 days in this group (target variable).

[0476] In this embodiment 5, the final modeling results are described by taking five features as examples: the current monthly credit repayment rate, the current remaining available credit limit of the credit card, the number of months since the credit card limit utilization rate in the past 12 months>90%, the average number of cash withdrawals from the credit card in the past 3 months, and the maximum asset size value in the past 3 months. In the feature conversion step, the five modeling variables have been converted in a corresponding form according to the correlation between the feature and the target variable. In the credit default probability modeling step, the five features screened by the feature fine screening step are substituted into the Sigmoid function (using the LOGISTIC process of SAS software) for logistic regression to calculate the credit default probability model.

[0477] The current monthly loan repayment rate reflects the current repayment situation of the borrower. For customers with an account age of less than 6 years, the higher the current repayment rate, the better the asset situation, the stronger the debt repayment ability, and the lower the possibility of serious overdue payments. Conversely, the higher the possibility of serious overdue payments. The conversion method is the original value.

[0478] The current remaining available credit limit of a credit card reflects the borrower's credit demand. For customers with an account age of less than 6 years, the lower the current remaining available credit limit of a credit card, the higher the borrower's recent credit demand, the greater the financial pressure, and the higher the possibility of a serious default; conversely, the higher the current remaining available credit limit of a credit card, the lower the borrower's financial pressure, and the higher the possibility of a serious default. The conversion method is the natural logarithm conversion in the continuous conversion.

[0479] The number of months since the credit card credit limit utilization rate in the past 12 months was > 90% reflects the borrower's credit limit utilization rate in the past year. For customers with an account age of less than 6 years, a credit limit utilization rate greater than 90% indicates a high demand for credit. Therefore, the smaller the number of months since the credit limit utilization rate is greater than 90%, the higher the customer's recent credit demand and the greater the possibility of serious overdue payments. Conversely, the smaller the possibility of serious overdue payments. The conversion method is cube root conversion.

[0480] The average number of cash withdrawals from credit cards in the past three months reflects the borrower's recent capital shortage. For customers with an account age of less than 6 years, the higher the average number of cash withdrawals, the worse the borrower's recent asset status, the more serious the capital shortage, and the greater the possibility of serious overdue payments. Conversely, the possibility of serious overdue payments is smaller. The conversion method is woe conversion value.

[0481] The maximum asset size in the past three months reflects the borrower's debt repayment ability. For customers with an account age of less than 6, the lower the maximum asset size in the past three months, the lower the borrower's debt repayment ability and the higher the possibility of a serious default; conversely, the higher the maximum asset size in the past three months, the stronger the borrower's debt repayment ability and the lower the possibility of a serious default. The conversion method is the square root conversion value.

[0482] In the step of calculating the probability of credit default, based on the data and coefficients obtained in the feature conversion step, the following model formula is used to predict the probability (P) that the borrower is in default:

[0483]

[0484] Where k is the number of features entering the model, and in Formula 1, k is 5.

[0485] α is the intercept term, with a range of values ​​from (-0.58156, -0.82194) and an optimal value of -0.701748; β1 is the coefficient corresponding to the current monthly credit repayment rate, with a range of values ​​from (-0.68379, -0.78681) and an optimal value of -0.7353; β2 is the coefficient corresponding to the current remaining available credit limit of the credit card, with a range of values ​​from (-0.60093, -0.66961) and an optimal value of -0.635274; β3 is the coefficient corresponding to the number of months from now when the credit card credit limit utilization rate in the past 12 months is > 90%. The value range is (-0.29231, -0.60137), and the optimal value is -0.446841; β4 is the coefficient corresponding to the average number of cash withdrawals of credit cards in the past three months, the value range is (-0.6339, -0.73691), and the optimal value is -0.685404; β5 is the coefficient corresponding to the maximum value of asset size in the past three months, the value range is (-0.08642, -0.22378), and the optimal value is -0.155097 (Note: The value range comes from the 95% confidence interval, that is, the 95% CI in the table below)

[0486] x1 is the original value of the current monthly credit repayment rate generated by the feature conversion step; x2 is the natural logarithm conversion value of the current remaining available credit limit of the credit card generated by the feature conversion step; x3 is the cube root conversion value of the number of months from now when the credit limit utilization rate of the credit card in the past 12 months is >90% generated by the feature conversion step; x4 is the woe conversion value of the average number of cash withdrawals of the credit card in the past 3 months generated by the feature conversion step; x5 is the square root conversion value of the maximum asset size value in the past 3 months generated by the feature conversion step. The model performance of some features is shown in Table 2 below.

[0487] Table 2

[0488] The P values ​​of all model characteristics are less than 0.05, indicating that the above characteristics are significantly correlated with default performance.

[0489]

[0490] Example 6 Construction of Alphad 2 Model

[0491] For Alphad 2 as a sub-model of the segmentation model, such as Figure 2 As shown in A, the above decision tree method is adopted, and the sample size obtained by classification is about 1 million. Its customer base is mainly current overdue customers with an account age greater than or equal to 6. The model is built to predict the probability of credit overdue for more than 90 days in this group (target variable).

[0492] In this embodiment 6, the final confirmation is to describe the final modeling results by taking six features as examples: the number of months from the minimum balance of the deposit account in the past 12 months, the number of times the credit was overdue > 0 in the past 6 months, the current cash withdrawal balance of the credit card (including cash installments) / current credit limit percentage, the current monthly credit limit utilization rate of the revolving loan, and the number of months with interest > 0 in the past 12 months of the special installment. In the feature conversion step, the conversion method selection has been carried out on the 5 modeling variables in the corresponding form according to the correlation between the feature and the target variable. In the credit default probability modeling step, the 5 features screened by the feature fine screening step are substituted into the Sigmoid function (using the LOGISTIC process of SAS software) for logistic regression to calculate the credit default probability model.

[0493] The number of months from the minimum balance of the deposit account in the past 12 months to the present reflects the change of the customer's assets. For customers with an account age of 6 or more and currently overdue, the closer the minimum balance in the past year is to the present, the worse the recent asset level is, and the greater the possibility of serious overdue, and vice versa. The possibility of serious overdue is smaller. The conversion method is cube root conversion.

[0494] The number of months with a repayment rate > 80% in the past three months reflects the customer's recent repayment situation. For customers with an age of 6 or more and currently overdue, the more months with a repayment rate > 80% in the past three months, the stronger the recent repayment ability and willingness, and the lower the possibility of serious overdue, and vice versa. The possibility of overdue is greater. The conversion method is the original value.

[0495] The number of times the credit is overdue > 0 in the past 6 months reflects the customer's historical overdue situation. For customers with an age greater than or equal to 6 and currently overdue, the more times the overdue is > 0 in the past 6 months, the worse the repayment behavior and debt repayment ability in the past period of time, and the greater the possibility of continuous serious overdue. Conversely, the possibility of continuous serious overdue is smaller. The conversion method is the original value.

[0496] The percentage of the current cash withdrawal balance (including cash installments) / current credit limit reflects the customer's current credit usage. For customers with an account age of 6 or more and currently overdue, the higher the proportion of the cash withdrawal balance (including cash installments) in the current credit limit, the more serious the current capital shortage, the worse the debt repayment ability, and the greater the possibility of serious overdue. Conversely, the possibility of serious overdue is smaller. The conversion form is the original value.

[0497] The current monthly credit utilization rate of revolving loans reflects the customer's current credit usage. For customers with an account age of 6 or more and currently overdue, the higher the current credit utilization rate, the greater the credit demand, the greater the debt repayment pressure, and the greater the possibility of serious overdue payments. Conversely, the possibility of serious overdue payments decreases. The conversion method is the original value.

[0498] The number of months with interest > 0 for special installment in the past 12 months reflects the repayment status of the customer's special installment business. For customers with an age of 6 or more and currently overdue, the larger the installment amount, the greater the interest, the greater the repayment pressure, and the greater the possibility of serious overdue. Conversely, the possibility of serious overdue is smaller. The conversion method is square root conversion.

[0499] In the step of calculating the probability of credit default, based on the data and coefficients obtained in the feature conversion step, the following model formula is used to predict the probability (P) that the borrower is in default:

[0500]

[0501] Where k is the number of features entering the model, and in Formula 2, k is 6.

[0502] α is the intercept term, with a range of values ​​from (-0.14511, -0.18322), and the optimal value is -0.164164; β1 is the coefficient corresponding to the number of months from the minimum balance of the deposit account in the past 12 months, with a range of values ​​from (0.00429, -0.06312), and the optimal value is -0.029414; β2 is the coefficient corresponding to the number of months with a credit repayment rate > 80% in the past 3 months, with a range of values ​​from (0.26668, 0.12932), and the optimal value is 0.198; β3 is the coefficient corresponding to the number of times the credit was overdue > 0 in the past 6 months, with a range of values ​​from (-0.48885, -0. 55331), the optimal value is -0.52108; β4 is the coefficient corresponding to the current cash withdrawal balance (including cash installment) of the credit card / the current credit limit percentage, the value range is (-0.51312, -0.51864), the optimal value is -0.515879; β5 is the coefficient corresponding to the utilization rate of the current monthly credit limit of the revolving loan, the value range is (-0.32363, -0.35507), the optimal value is -0.339353; β6 is the coefficient corresponding to the number of months with interest > 0 in the past 12 months of the special installment, the value range is (-0.38218, -0.63469), the optimal value is -0.508438. (Note: The value range comes from the 95% confidence interval, that is, the 95% CI in the table below)

[0503] x1 is the cube root conversion value of the minimum balance of the deposit account in the past 12 months generated by the feature conversion step; x2 is the original value of the number of months with a credit repayment rate > 80% in the past 3 months generated by the feature conversion step; x3 is the original value of the number of credit overdue > 0 in the past 6 months generated by the feature conversion step; x4 is the natural logarithm conversion value of the current cash withdrawal balance (including cash installments) of the credit card / current credit limit percentage generated by the feature conversion step; x5 is the original value of the current monthly credit limit utilization rate of the revolving loan generated by the feature conversion step; x6 is the square root conversion value of the number of months with interest > 0 in the past 12 months of the special installment generated by the feature conversion step. The model performance of some features is shown in Table 3 below:

[0504] Table 3

[0505] The P values ​​of all model characteristics are less than 0.05, indicating that the above characteristics are significantly correlated with default performance.

[0506]

[0507] Example 7 Construction of Alphad3 model

[0508] For Alphad 3 as a sub-model of the segmentation model, such as Figure 2 As shown in A, by adopting the above decision tree method, the sample size obtained by classification is about 360,000. Its customer base is mainly customers who have applied for credit business and whose current credit limit utilization rate is greater than or equal to 100%. The model is built to predict the probability of credit overdue for more than 90 days in this group (target variable).

[0509] In this embodiment 7, the final confirmation is to use six features as examples to describe the final modeling results: the number of months with the largest continuous increase in the asset size value month-on-month in the past 3 months, the percentage of overdue interest on mortgage loans in the past 6 months to all repayment amounts in the past 6 months, the percentage of months with a credit limit utilization rate of >70% in the past 3 months, the minimum credit card repayment rate in the past 6 months, the number of months since the credit card interest >0 in the past 3 months, and the total amount of cash withdrawals for credit card consumption in the current month (including installments) / the total amount of cash withdrawals for consumption in the previous month (including installments). In the feature conversion step, the conversion method selection has been carried out on the 6 modeling variables in the corresponding form according to the correlation between the feature and the target variable. In the credit default probability modeling step, the 6 features screened by the feature fine screening step are substituted into the Sigmoid function (using the LOGISTIC process of SAS software) to perform logistic regression to calculate the credit default probability model.

[0510] The number of months with the largest consecutive increase in asset size over the past three months reflects the recent changes in the customer's asset level. For customers who are not currently overdue and whose credit limit utilization rate is greater than or equal to 100%, the greater the number of consecutive months of asset size increase, the better the asset level, and the more assets available for repayment, the lower the possibility of serious overdue, and vice versa. The conversion method is natural logarithm conversion.

[0511] The percentage of overdue interest on mortgages in the past six months to all repayments in the past six months reflects the severity of overdue mortgages held by customers in the past six months. For customers who are not currently overdue and whose credit limit utilization rate is greater than or equal to 100%, the higher the percentage of overdue interest in the past six months to all repayments, the worse the debt repayment ability is and the greater the possibility of serious overdue, and vice versa. The possibility of serious overdue is smaller. The conversion method is the original value.

[0512] The percentage of months with a credit line utilization rate of >70% in the past three months reflects the customer's recent credit demand. For customers who are not currently overdue and whose credit line utilization rate is greater than or equal to 100%, the higher the percentage of months with a credit line utilization rate of more than 70%, the more urgent the recent credit demand, the greater the debt repayment pressure, and the greater the possibility of serious overdue. Conversely, the possibility of serious overdue is smaller. The conversion method is cube root conversion.

[0513] The minimum credit card repayment rate in the past 6 months reflects the customer's repayment ability and repayment behavior in the past period of time. For customers who are not currently overdue and whose credit limit utilization rate is greater than or equal to 100%, the smaller the minimum credit card repayment rate in the past 6 months, the worse the customer's economic level and debt repayment ability, and the higher the possibility of serious overdue, and vice versa. The possibility of serious overdue is smaller. The conversion method is dummy variable conversion.

[0514] The number of months since the credit card interest rate was >0 in the past three months reflects the customer's recent failure to pay back the full amount of the credit card. For customers who are not currently overdue and whose credit limit utilization rate is greater than or equal to 100%, the smaller the number of months since the credit card interest rate was >0, the lower the recent repayment ability, the greater the debt repayment pressure, and the greater the possibility of serious overdue, and vice versa. The possibility of serious overdue will decrease. The conversion method is natural logarithm conversion.

[0515] The percentage of the total amount of cash withdrawals from credit card consumption in the current month (including installments) / the total amount of cash withdrawals from consumption in the previous month (including installments) reflects the customer's recent credit card cash withdrawal usage. For customers who are not currently overdue and whose credit limit utilization rate is greater than or equal to 100%, the larger the ratio of cash withdrawals in the current month compared to the previous month, the tighter the current funds are, the greater the debt repayment ability is, and the possibility of serious overdue payments increases accordingly. Conversely, the possibility of serious overdue payments decreases accordingly. The conversion method is WOE conversion.

[0516] In the step of calculating the probability of credit default, based on the data and coefficients obtained in the feature conversion step, the following model formula is used to predict the probability (P) that the borrower is in default:

[0517]

[0518] Where k is the number of features entering the model, and in Formula 3, k is 6.

[0519] α is the intercept term, with a range of values ​​of (-0.11402, -0.35754), and the optimal value is -0.2357787; β1 is the coefficient corresponding to the maximum number of consecutive months of increase in asset size in the past three months, with a range of values ​​of (-0.19897, -0.31636), and the optimal value is -0.2576679; β2 is the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months, with a range of values ​​of (-0.03843, -0.28001), and the optimal value is -0.1592187; β3 is the coefficient corresponding to the proportion of months with a credit limit utilization rate of > 70% in the past three months, with a range of values ​​of (-0.71885, -0.91682), the optimal value is -0.8178348; β4 is the coefficient corresponding to the minimum credit card repayment rate in the past 6 months, the value range is (-0.18116, -0.49436), the optimal value is -0.3377601; β5 is the coefficient corresponding to the number of months from now when the credit card interest is > 0 in the past 3 months, the value range is (-0.22644, -0.47791), the optimal value is -0.352176; β6 is the coefficient corresponding to the percentage of the total amount of cash withdrawals for credit card consumption in the current month (including installments) / the total amount of cash withdrawals for consumption in the previous month (including installments), the value range is (-0.05659, -0.38296), the optimal value is -0.2197794. (Note: The numerical range comes from the 95% confidence interval, that is, the 95% CI in the table below)

[0520] x1 is the natural logarithm conversion value of the maximum number of consecutive months of month-on-month increase in asset size values ​​in the past three months generated by the feature conversion step; x2 is the original value of the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months generated by the feature conversion step; x3 is the cube root conversion value of the number of months with a credit limit utilization rate > 70% in the past three months generated by the feature conversion step; x4 is the dummy variable conversion value of the minimum credit card repayment rate in the past six months generated by the feature conversion step; x5 is the natural logarithm conversion value of the number of months since the credit card interest rate > 0 in the past three months generated by the feature conversion step; x6 is the WOE conversion value of the total amount of credit card cash withdrawals (including installments) in the current month / total amount of cash withdrawals (including installments) in the previous month generated by the feature conversion step.

[0521] The model performance of some features is shown in Table 4 below:

[0522] Table 4

[0523]

[0524] The P values ​​of all model characteristics are less than 0.05, indicating that the above characteristics are significantly correlated with default performance.

[0525] Example 8 Alphad 4 model construction

[0526] For Alphad 4 as a sub-model of the subdivision model, such as Figure 2 As shown in A, the sample size obtained by classification using the above decision tree method is about 3 million. Its customer base is mainly customers who have applied for credit business and are not currently overdue and whose credit limit utilization rate is less than or equal to 100%. The model is built to predict the probability of credit overdue for more than 90 days in this group (target variable).

[0527] In Example 8, it is finally confirmed that the final modeling results are described by taking the current deposit balance, the average balance of investment and financial management accounts in the past 12 months, the proportion of the largest continuous increase in the asset size value in the past 3 months, the current remaining available credit card limit, the proportion of the number of months with a credit card repayment rate of <90% in the past 12 months, the maximum number of consecutive increase in the credit card consumption balance (including consumption installments) in the past 3 months, and the number of months of credit card cycles and installments in the past 6 months. In the feature conversion step, the conversion method selection has been carried out on the 7 modeling variables according to the correlation between the feature and the target variable. In the credit default probability modeling step, the 7 features screened by the feature fine screening step are substituted into the Sigmoid function (using the LOGISTIC process of SAS software) to perform logistic regression to calculate the credit default probability model.

[0528] The current deposit balance reflects the borrower's ability to repay. For customers who are not overdue and whose credit line utilization rate is less than or equal to 100%, the lower the current deposit balance, the lower the borrower's ability to repay, and the higher the possibility of a serious default; conversely, the possibility of a serious default will also decrease. The conversion method is natural logarithm.

[0529] The average balance of investment and financial management accounts in the past 12 months reflects the average level of the customer's financial management assets in the past period of time. For customers who are not currently overdue and whose credit line utilization rate is less than or equal to 100%, the higher the average investment and financial management balance in the past year, the better the customer's financial management habits and ability, the better the overall asset level, the better the debt repayment ability, and the smaller the possibility of serious overdue in the future. Conversely, the greater the possibility of serious overdue in the future. The conversion method is natural logarithm.

[0530] The percentage of the largest consecutive increase in asset size in the past three months reflects the recent changes in customers' assets. For customers who are not currently overdue and whose credit line utilization rate is less than or equal to 100%, the more consecutive months of asset increase in the past three months, the better the recent asset level, the stronger the debt repayment ability, and the lower the possibility of serious overdue, and vice versa. The conversion method is the original value.

[0531] The current remaining available credit limit of a credit card reflects the borrower's credit demand. For customers who are not currently overdue and whose credit limit utilization rate is less than or equal to 100%, the lower the current remaining available credit limit of a credit card, the greater the borrower's recent credit demand may be, resulting in greater financial pressure and a lower possibility of serious default; conversely, the higher the current remaining available credit limit of a credit card, the less financial pressure the borrower has, and the higher the possibility of serious default. The conversion method is the square root conversion in the continuous conversion.

[0532] The percentage of months with a credit card repayment rate of <90% in the past 12 months reflects the borrower's credit card repayment situation in the past period of time. For customers who are not currently overdue and whose credit line utilization rate is less than or equal to 100%, the fewer the percentage of months with a credit card repayment rate of <90% in the past 12 months, the stronger the customer's repayment habits and repayment ability, the higher the credit management level, and the lower the possibility of serious overdue payments. Conversely, the greater the possibility of serious overdue payments. The conversion method is the original value.

[0533] The maximum number of consecutive months of credit card consumption balance (including consumption installments) in the past three months reflects the customer's recent credit situation. For customers who are not currently overdue and whose credit limit utilization rate is less than or equal to 100, the more consecutive months of recent consumption balance increase, the greater the recent credit demand, the greater the debt repayment pressure, and the greater the possibility of serious overdue. Conversely, the smaller the possibility of serious overdue. The conversion method is dummy variable conversion.

[0534] The number of months of credit card revolving and installment use in the past six months reflects the customer's credit demand and use in the past period of time. For customers who are not currently overdue and whose credit limit utilization rate is less than or equal to 100, the more months of revolving and installment use in the past six months, the higher the credit demand, the worse the repayment ability, and the greater the possibility of serious overdue, and vice versa. The possibility of serious overdue is smaller. The conversion method is cube root conversion.

[0535] In the step of calculating the probability of credit default, based on the data and coefficients obtained in the feature conversion step, the following model formula is used to predict the probability (P) that the borrower is in default:

[0536]

[0537] Where k is the number of features entering the model. In Formula 4, k is 7.

[0538] α is the intercept term, with a range of values ​​from (-0.12673, -0.19974), and the optimal value is -0.1632328; β1 is the coefficient corresponding to the deposit balance at the current point in time, with a range of values ​​from (-0.44362, -0.53469), and the optimal value is -0.4891512; β2 is the coefficient corresponding to the average balance of investment and financial management accounts in the past 12 months, with a range of values ​​from (-0.181, -0.52118), and the optimal value is -0.3510896; β3 is the coefficient corresponding to the proportion of the largest consecutive increase in the asset size value in the past three months, with a range of values ​​from (0.77324, 0.69876), and the optimal value is 0.736; β4 is the current remaining available credit card balance The corresponding coefficient of the credit limit has a range of values ​​from (-0.16337, -0.27663), and the optimal value is -0.22; β5 is the corresponding coefficient of the proportion of months with a credit card repayment rate of <90% in the past 12 months, with a range of values ​​from (0.1674, 0.0106), and the optimal value is 0.089; β6 is the corresponding coefficient of the maximum number of consecutive months of credit card consumption balance (including consumption installments) in the past 3 months, with a range of values ​​from (-0.57383, -0.78447), and the optimal value is -0.6791512; β7 is the corresponding coefficient of the number of months of credit card cycle and installment use in the past 6 months, with a range of values ​​from (-0.01967, -0.06671), and the optimal value is -0.0431908. (Note: The numerical range comes from the 95% confidence interval, i.e., the 95% CI in the table below)

[0539] x1 is the natural logarithm conversion value of the deposit balance at the current time point generated by the feature conversion step; x2 is the natural logarithm conversion value of the average balance of the investment and financial management account in the past 12 months generated by the feature conversion step; x3 is the original value of the proportion of the largest consecutive increase in the asset size value in the past 3 months generated by the feature conversion step; x4 is the square root conversion value of the current remaining available credit card limit generated by the feature conversion step; x5 is the original value of the proportion of the number of months with a credit card repayment rate <90% in the past 12 months generated by the feature conversion step; x6 is the dummy variable conversion value of the maximum consecutive increase in the number of months of credit card consumption balance (including consumption installments) in the past 3 months generated by the feature conversion step; x7 is the cube root conversion value of the number of months of credit card cycle and installment use in the past 6 months generated by the feature conversion step.

[0540] The model performance of some features is shown in Table 5 below:

[0541] Table 5

[0542]

[0543] The P values ​​of all model characteristics are less than 0.05, indicating that the above characteristics are significantly correlated with default performance.

[0544] Example 9 Alphad 5 model construction

[0545] For Alphad 5 as a sub-model of the segmentation model, such as Figure 2 As shown in Figure B, using the above decision tree method, the sample size obtained for classification is about 20,000. The customer base is mainly new credit applicants. The model is built to predict the probability of credit overdue for more than 90 days in this group (target variable).

[0546] In Example 9, it is finally confirmed that the final modeling results are described by taking the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, the maximum salary in the past 12 months, the maximum balance of the investment and financial management account in the past 6 months, and the minimum balance of the deposit account at a point in time in the past 6 months. In the selection of the conversion method in the feature conversion step, the four modeling variables have been converted in a corresponding form according to the correlation between the feature and the target variable. In the credit default probability modeling step, the four features screened by the feature fine screening step are substituted into the Sigmoid function (using the LOGISTIC process of SAS software) for logistic regression to calculate the credit default probability model.

[0547] The average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months reflects the changes in the borrower's asset level. For customers who have recently applied, the increase in their asset level means that the asset size is constantly increasing. The larger the ratio of the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, the better the borrower's recent asset level and the more stable the financial situation, and the lower the possibility of default. The smaller the ratio of the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, the worse the borrower's recent asset level and the more unstable the financial level, and the higher the possibility of default. The conversion method is the original value.

[0548] The maximum salary in the past 12 months reflects the income level of the customer in the past period of time. For customers who have recently applied, the higher their average income level, the more assets they can use to repay credit debts, and the lower the possibility of serious overdue payments. Conversely, the higher the possibility of serious overdue payments. The conversion method is natural logarithm conversion.

[0549] The maximum balance of the investment and financial management account in the past 6 months reflects the customer's recent investment and financial management level. For customers who have recently applied, the larger the balance of the investment and financial management account in the past period of time, the stronger the financial management ability, the better the asset quality, the stronger the debt repayment ability, and the lower the possibility of serious overdue payments, and vice versa. The possibility of serious overdue payments increases. The conversion method is natural logarithm conversion.

[0550] The minimum balance of the deposit account in the past six months reflects the deposit situation of the borrower in recent years. For customers who have recently applied, the minimum deposit amount can reflect their asset quality. The higher the minimum balance of the deposit account in the past six months, the more abundant the borrower's funds are, and the lower the possibility of serious default; conversely, the lower the minimum balance of the deposit account in the past six months, the less funds the borrower has, and the higher the possibility of serious default. The conversion method is natural logarithm conversion.

[0551] In the step of calculating the probability of credit default, based on the data and coefficients obtained in the feature conversion step, the following model formula is used to predict the probability (P) that the borrower is in default:

[0552]

[0553] Where k is the number of features entering the model. In Formula 5, k is 4.

[0554] α is the intercept term, with a range of values ​​of (-2.04845, -2.31097), and the optimal value is -2.1797076; β1 is the corresponding coefficient of the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, with a range of values ​​of (-0.20978, -0.34796), and the optimal value is -0.2788716; β2 is the corresponding coefficient of the maximum salary in the past 12 months, with a range of values ​​of (-0.055 51,-0.39342), the optimal value is -0.2244648; β3 is the coefficient corresponding to the maximum balance of the investment and financial management account in the past 6 months, the value range is (-0.1207,-0.12475), the optimal value is -0.122724; β4 is the coefficient corresponding to the minimum balance of the deposit account in the past 6 months, the value range is (-0.68315,-0.93368), the optimal value is -0.808416.

[0555] x1 is the original value of the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months generated by the feature conversion step; x2 is the natural logarithm transformation value of the maximum salary in the past 12 months generated by the feature conversion step; x3 is the dummy variable transformation value of the maximum balance of the investment and financial management account in the past 6 months generated by the feature conversion step; x4 is the natural logarithm transformation value of the minimum balance of the deposit account at a certain point in time in the past 6 months generated by the feature conversion step.

[0556] The model performance of some features is shown in Table 6 below:

[0557] Table 6

[0558]

[0559] The P values ​​of all model characteristics are less than 0.05, indicating that the above characteristics are significantly correlated with default performance.

[0560] Example 10

[0561] The P values ​​calculated by the above formulas can be further used to calculate the score of any customer.

[0562] The scoring calculation step is used to convert the calculated credit default probability into a score of 0-1000 using a pre-stored default score conversion code.

[0563] In the scoring calculation module, the following formula is used to calculate and generate the credit score used to characterize the borrower:

[0564]

[0565]

[0566] Among them, P is the borrower's default probability (P) generated in the credit default probability calculation module, A is 54.2458; B is 115.4156, and the round function rounds the calculated score to the integer value; finally, the score greater than 1000 is set to 1000 points, and the score less than 0 is set to 0 points.

[0567] In this embodiment, the P value calculated in the above embodiments can be substituted to calculate the score.

[0568] The KS (Kolmogorov-Smirnov) statistic was proposed by two Soviet mathematicians, ANKolmogorov and NVSmirnov. In risk control, KS is often used to evaluate the discrimination of models. The greater the discrimination, the stronger the risk ranking ability of the model.

[0569] The KS statistic is based on the empirical cumulative distribution function (ECDF), which is generally defined as:

[0570] KS=max(|cum(bad_rate)-cum(good_rate)|)

[0571] According to the prediction model, the KS curve is drawn. The KS of this embodiment is 80.77. It can be seen that the model constructed in this embodiment performs well in assessing customer credit risk. Figure 1 shown.

[0572] The following shows the results of the KS values ​​of the model applied to the development sample and the reserved validation sample in the development sample and the validation sample. From Table 7, it can be seen that the model has a good discrimination effect in the training set and the validation set (where the sample size ratio of the training set and the validation set is 6:4) and the overall sample, that is, the discrimination effect is very good in all sub-models.

[0573] Table 7

[0574]

Claims

1. A method for calculating retail credit risk, comprising: A data collection step, which obtains retail credit prediction data of the sample to be predicted; A step of classifying samples to be predicted, which classifies the samples to be predicted based on a decision tree method to determine a sub-model for calculating the probability of credit default; The credit default probability calculation step is to substitute the retail credit prediction data into the credit default probability sub-model to calculate the credit default probability of the sample to be predicted, and preferably the credit default probability is the credit default probability.

2. The method according to claim 1, further comprising: After calculating the credit default probability, the step of calculating the credit score of the sample to be predicted is used to calibrate the calculated credit default probability to a standardized score of 0-1000 points.

3. The method according to claim 1 or 2, wherein: The retail credit prediction data includes original retail credit prediction data of the sample to be predicted and derived retail credit prediction data processed based on the original retail credit prediction data; Preferably, the original retail credit prediction data includes: Credit card basic data, which is based on all available data during the sample user's credit card creation and usage process. Basic data on personal loans, which is all available data based on the loan application and usage behavior of sample users. Basic customer information data, which is based on the attributes of the sample users themselves but is not directly related to their behavior in financial institutions. Basic data on personal financial assets, which includes all other financial assets and financial transaction data of sample users in financial institutions that are not related to credit cards and loans.

4. The method according to any one of claims 1 to 3, wherein: The derived retail credit prediction data processed based on the original retail credit prediction data refers to the data obtained by processing the collected original retail credit prediction data based on the time dimension, space dimension, frequency dimension and statistical information dimension; Preferably, the derived retail credit forecast data includes but is not limited to: Derived retail credit forecast data obtained by processing based on sample relationship length, Derivative retail credit forecast data obtained by processing time interval variables, Derivative retail credit forecast data obtained by processing based on the frequency of sample behavior, Derivative retail credit forecast data obtained by processing the current time point of the sample, Derivative retail credit forecast data obtained by processing based on sample continuous behavior, Derived retail credit forecast data obtained by processing sample data based on statistical information dimensions.

5. The method according to any one of claims 1 to 4, wherein: The retail credit prediction data is selected from one, two, three, four, five, six or seven of the following: The current monthly credit repayment rate, the current remaining available credit limit of the credit card, the number of months since the credit card credit limit utilization rate in the past 12 months was >90%, the average number of cash withdrawals from the credit card in the past 3 months, the maximum asset size value in the past 3 months, the number of months since the minimum balance of the deposit account in the past 12 months, the number of months in which the credit repayment rate in the past 3 months was >80%, the number of credit overdue payments >0 in the past 6 months, the current credit card cash withdrawal balance (including cash installments) / current credit limit percentage, the current monthly credit limit utilization rate of revolving loans, the corresponding coefficient of the number of months with interest >0 for special installments in the past 12 months, the number of consecutive months with the largest month-on-month increase in asset size in the past 3 months, the percentage of overdue interest on mortgage loans in the past 6 months to all repayment amounts in the past 6 months, the proportion of months with credit limit utilization rate >70% in the past 3 months, the past 6 Minimum monthly credit card repayment rate, number of months since the credit card interest rate was >0 in the past 3 months, percentage of total cash withdrawal amount of credit card consumption in the current month (including installments) / total cash withdrawal amount of consumption in the previous month (including installments), current deposit balance, average balance of investment and financial management accounts in the past 12 months, percentage of the largest consecutive increase in asset size in the past 3 months, current remaining available credit card limit, percentage of months with credit card repayment rate <90% in the past 12 months, number of months with the largest consecutive increase in credit card consumption balance (including consumption installments) in the past 3 months, number of months of credit card revolving and installment use in the past 6 months, average monthly asset size in the past 6 months / average monthly asset size in the past 12 months, maximum salary in the past 12 months, maximum balance of investment and financial management accounts in the past 6 months, minimum balance of deposit accounts in the past 6 months.

6. The method according to any one of claims 1 to 5, wherein: The step of classifying the sample to be predicted includes the following sub-steps: Whether the sample to be tested is a customer who has applied for credit business from a financial institution; Whether the sample to be tested has an age greater than or equal to G1; Whether the sample to be tested is a customer with overdue credit; Whether the current credit utilization rate of the sample to be tested is less than a%; Based on the above sub-steps, the samples to be predicted are classified to determine the sub-model for calculating the probability of credit default. Under the premise of ensuring the rationality of business logic, the order of the above sub-steps can be set arbitrarily; It is preferred to classify the predicted samples in the following order: First, determine whether the sample to be tested is a customer who has applied for credit business at a financial institution; Then determine whether the sample to be tested has an account age greater than or equal to G1; Then, determine whether the credit of the sample to be tested is overdue; Then determine whether the current credit limit utilization rate of the sample to be tested is less than a%.

7. The method according to any one of claims 1 to 6, wherein: After performing feature conversion on the retail credit prediction data, the data is substituted into the credit default probability sub-model to calculate the credit default probability of the sample to be predicted. The feature conversion step includes: Based on the feature type of the retail credit prediction data that needs to be substituted into the credit default probability sub-model, the WOE method, dummy feature method or continuous method is selected for feature conversion.

8. The method according to claim 7, wherein: The continuous method for feature conversion includes the following methods: directly selecting the original value, calculating the square of the original data, calculating the square root of the original data, calculating the cube root of the original data, or calculating the natural logarithm of the original data to perform continuous feature conversion.

9. The method according to any one of claims 1 to 8, wherein: The credit default probability sub-model is a model constructed based on sample retail credit prediction data and credit default probability using logistic regression based on the existing user population.

10. The method according to any one of claims 1 to 9, wherein: The retail credit forecast data is selected from: the current monthly credit repayment rate, the current remaining available credit limit of the credit card, the number of months since the credit card credit limit utilization rate > 90% in the past 12 months, the average number of cash withdrawals from the credit card in the past three months, and one, two, three, four or five of the maximum asset size values ​​in the past three months.

11. The method according to claim 10, wherein: The steps to calculate the probability of credit default include: The retail credit prediction data is selected from: the current monthly credit repayment rate, the current remaining available credit limit of the credit card, the number of months since the credit limit utilization rate of the credit card in the past 12 months was > 90%, the average number of cash withdrawals from the credit card in the past three months, and the maximum value of the asset size in the past three months; Preferably, The current monthly repayment rate of the credit card is converted in a continuous manner; the current remaining available credit limit of the credit card is converted in a continuous manner; the number of months since the credit limit utilization rate of the credit card in the past 12 months was >90% is converted in a continuous manner; the average number of cash withdrawals from the credit card in the past three months is converted in a WOE manner; the maximum value of the asset size in the past three months is converted in a continuous manner; More preferably, The current monthly credit repayment rate is converted into the original value calculation method using a continuous method; the current remaining available credit limit of the credit card is converted into the natural logarithm calculation method using a continuous method; the number of months from now when the credit limit utilization rate of the credit card in the past 12 months is >90% is converted into the cube root calculation method using a continuous method; the maximum asset size value in the past 3 months is converted into the square root calculation method using a continuous method.

12. The method according to claim 11, wherein: The converted values ​​of the five features, namely, the current monthly credit repayment rate, the current remaining available credit limit of the credit card, the number of months since the credit card limit utilization rate > 90% in the past 12 months, the average number of cash withdrawals from the credit card in the past three months, and the maximum asset size value in the past three months, are substituted into the sub-model constructed by logistic regression based on the sample retail credit prediction data and credit default probability to calculate the default probability of the sample to be predicted.

13. The method according to claim 12, wherein: The sub-model is shown in the following formula 1: Where k is the number of features entering the model, preferably k is 5, α is the intercept term, the value range is (-0.58156, -0.82194), and the optimal value is -0.701748; β1 is the coefficient corresponding to the current monthly loan repayment rate, with a value range of (-0.68379, -0.78681), and the optimal value is -0.7353; β2 is the coefficient corresponding to the current remaining available credit limit of the credit card, with a value range of (-0.60093, -0.66961), and the optimal value is -0.635274; β3 is the coefficient of the number of months from now when the credit limit utilization rate of the credit card in the past 12 months is >90%, with a value range of (-0.29231, -0.60137) and the optimal value is -0.446841; β4 is the coefficient corresponding to the average number of cash withdrawals of credit cards in the past three months, with a value range of (-0.6339, -0.73691), and the optimal value is -0.685404; β5 is the coefficient corresponding to the maximum value of asset size in the past three months, with a value range of (-0.08642, -0.22378) and an optimal value of -0.155097; x1 is the original value of the current monthly credit repayment rate generated by the feature conversion step; x2 is the natural logarithm transformation value of the current remaining available credit limit of the credit card generated in the feature conversion step; x3 is the cube root conversion value of the number of months from now when the credit card limit utilization rate in the past 12 months is >90% generated in the feature conversion step; x4 is the WOE conversion value of the average number of cash withdrawals from credit cards in the past three months generated in the feature conversion step; x5 is the square root transformed value of the maximum asset size value in the past three months generated in the feature transformation step.

14. The method according to claim 13, wherein: After calculating the credit default probability, the step of calculating the credit score of the sample to be predicted is to use the following formula to calculate and generate a credit score for characterizing the borrower: Among them, P is the borrower's default probability (P) generated in the credit default probability calculation module, A is 54.2458; B is 115.4156, and the round function rounds the calculated score to the integer value; finally, the score greater than 1000 is set to 1000 points, and the score less than 0 is set to 0 points.

15. The method according to any one of claims 1 to 9, wherein: The retail credit forecast data is selected from: the number of months from the minimum balance of the deposit account in the past 12 months, the number of months with a credit repayment rate > 80% in the past 3 months, the number of credit overdue payments > 0 in the past 6 months, the current cash withdrawal balance of the credit card (including cash installments) / current credit limit percentage, the current monthly credit limit utilization rate of revolving loans, and one, two, three, four, five or six corresponding coefficients of the number of months with special installment interest > 0 in the past 12 months.

16. The method according to claim 15, wherein: The steps to calculate the probability of credit default include: The number of months from the minimum balance of the deposit account in the past 12 months, the number of months with a credit repayment rate > 80% in the past 3 months, the number of credit overdue payments > 0 in the past 6 months, the current credit card withdrawal balance (including cash installments) / current credit limit percentage, the current monthly credit limit utilization rate of revolving loans, and the number of months with interest > 0 in the past 12 months for special installments are converted into features. Preferably, The number of months from the minimum balance of the deposit account in the past 12 months to the present is converted in a continuous manner; the number of months with a credit repayment rate > 80% in the past 3 months is converted in a continuous manner; the number of credit overdue payments > 0 in the past 6 months is converted in a continuous manner; the current credit card cash withdrawal balance (including cash installments) / current credit limit percentage is converted in a continuous manner; the current monthly credit limit utilization rate of revolving loans is converted in a continuous manner; the number of months with interest > 0 in the past 12 months of special installments is converted in a continuous manner; More preferably, The number of months from the minimum balance of the deposit account in the past 12 months to the present is converted in a continuous manner to the number of months from the present minimum balance of the deposit account in the past 12 months using the cube root calculation method; the number of months in which the credit repayment rate in the past 3 months was >80% is converted in a continuous manner to the number of months in which the credit repayment rate in the past 3 months was >80% using the original value calculation method; the number of credit overdue payments >0 in the past 6 months is converted in a continuous manner to the number of credit overdue payments >0 in the past 6 months using the original value calculation method; the current cash withdrawal balance (including cash installments) of the credit card / current credit limit percentage is converted in a continuous manner to the current credit card cash withdrawal balance (including cash installments) / current credit limit percentage using the natural logarithm calculation method; the current monthly credit limit utilization rate of the revolving loan is converted in a continuous manner to the original value calculation method; the number of months in which the interest on the special installment loan is >0 in the past 12 months is converted in a continuous manner to the number of months in which the interest on the special installment loan is >0 in the past 12 months using the square root calculation method.

17. The method according to claim 15, wherein: The converted values ​​of the six features, namely, the number of months from the present when the minimum balance of the deposit account in the past 12 months was from now, the number of months with a credit repayment rate > 80% in the past 3 months, the number of credit overdue payments > 0 in the past 6 months, the current cash withdrawal balance of the credit card (including cash installments) / current credit limit percentage, the current monthly credit limit utilization rate of revolving loans, and the number of months with special installment interest > 0 in the past 12 months, are substituted into the sub-model constructed by logistic regression based on the sample retail credit prediction data to calculate the default probability of the sample to be predicted.

18. The method according to claim 17, wherein: The sub-model is shown in the following formula 2: Where k is the number of features entering the model, and k is preferably 6; α is the intercept term, the value range is (-0.14511,-0.18322), and the optimal value is -0.164164; β1 is the coefficient corresponding to the number of months from the present when the minimum balance of the deposit account was the same in the past 12 months. The value range is (0.00429, -0.06312), and the optimal value is -0.029414; β2 is the coefficient corresponding to the number of months in which the loan repayment rate was >80% in the past three months, with a value range of (0.26668, 0.12932) and an optimal value of 0.198; β3 is the coefficient corresponding to the number of times the credit is overdue > 0 in the past 6 months, with a value range of (-0.48885, -0.55331) and an optimal value of -0.52108; β4 is the coefficient corresponding to the current cash withdrawal balance (including cash installment) of the credit card / current credit limit percentage, with a value range of (-0.51312, -0.51864), and the optimal value is -0.515879; β5 is the coefficient corresponding to the utilization rate of the current monthly quota of the revolving loan, with a value range of (-0.32363, -0.35507), and the optimal value is -0.339353; β6 is the coefficient corresponding to the number of months with interest > 0 in the past 12 months of the special installment plan, with a value range of (-0.38218, -0.63469) and an optimal value of -0.508438; x1 is the cube root conversion value of the minimum balance of the deposit account in the past 12 months generated by the feature conversion step. x2 is the original value of the number of months with a credit repayment rate > 80% in the past three months generated by the feature conversion step; x3 is the original value of the number of credit overdue > 0 in the past 6 months generated by the feature conversion step; x4 is the natural logarithm transformation value of the current cash withdrawal balance (including cash installments) of the credit card / current credit limit percentage generated in the feature conversion step; x5 is the original value of the current monthly credit limit utilization rate of the revolving loan generated in the feature conversion step; x6 is the square root conversion value of the number of months with interest > 0 in the past 12 months for the special installment generated in the feature conversion step.

19. The method according to claim 18, wherein: After calculating the credit default probability, the step of calculating the credit score of the sample to be predicted is to use the following formula to calculate and generate a credit score for characterizing the borrower: Among them, P is the borrower's default probability (P) generated in the credit default probability calculation module, A is 54.2458; B is 115.4156, and the round function rounds the calculated score to the integer value; finally, the score greater than 1000 is set to 1000 points, and the score less than 0 is set to 0 points.

20. The method according to any one of claims 1 to 9, wherein: The retail credit forecast data is selected from: the number of consecutive months with the largest month-on-month increase in asset size in the past three months, the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months, the percentage of months with credit limit utilization rate > 70% in the past three months, the minimum credit card repayment rate in the past six months, the number of months since the credit card interest rate > 0 in the past three months, and the percentage of total cash withdrawal amount of credit card consumption in the current month (including installments) / total cash withdrawal amount of consumption in the previous month (including installments).

21. The method according to claim 20, wherein: The steps to calculate the probability of credit default include: The features are converted into the maximum number of consecutive months of asset size increase month-on-month in the past three months, the percentage of overdue interest on mortgage loans in the past six months to all repayments in the past six months, the percentage of months with credit limit utilization rate > 70% in the past three months, the minimum credit card repayment rate in the past six months, the number of months since the credit card interest rate > 0 in the past three months, and the percentage of total cash withdrawal amount of credit card consumption in the current month (including installments) / total cash withdrawal amount of consumption in the previous month (including installments). Preferably, The continuous method is used to convert the number of months with the largest continuous increase in the asset size value month-on-month in the past three months; the continuous method is used to convert the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months; the continuous method is used to convert the proportion of months with credit limit utilization rate > 70% in the past three months; the dummy variable method is used to convert the minimum credit card repayment rate in the past six months; the continuous method is used to convert the number of months since the credit card interest rate > 0 in the past three months; the WOE method is used to convert the percentage of the total amount of cash withdrawals for credit card consumption in the current month (including installments) / the total amount of cash withdrawals for consumption in the previous month (including installments); More preferably, The continuous method is used to convert the number of months with the largest consecutive month-on-month increase in asset size over the past three months into the number of months with the largest consecutive month-on-month increase in asset size over the past three months using the natural logarithm calculation method; the continuous method is used to convert the percentage of overdue interest on mortgage loans in the past six months to the total repayment amount in the past six months into the percentage of overdue interest on mortgage loans in the past six months to the total repayment amount in the past six months using the original value calculation method; the continuous method is used to convert the proportion of months with credit limit utilization rate > 70% in the past three months into the proportion of months with credit limit utilization rate > 70% in the past three months using the cube root calculation method; the continuous method is used to convert the number of months from now when the credit card interest rate in the past three months is >0 into the number of months from now when the credit card interest rate in the past three months is >0 using the natural logarithm calculation method.

22. The method according to claim 20, wherein: The converted values ​​of six features, including the number of months with the largest month-on-month increase in asset size in the past three months, the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months, the percentage of months with credit limit utilization rate > 70% in the past three months, the minimum credit card repayment rate in the past six months, the number of months since the credit card interest rate > 0 in the past three months, and the percentage of total cash withdrawals for credit card consumption in the current month (including installments) / total cash withdrawals for consumption in the previous month (including installments), are substituted into the sub-model constructed by logistic regression based on the sample retail credit prediction data and credit default probability to calculate the default probability of the sample to be predicted.

23. The method according to claim 22, wherein: The sub-model is shown in Formula 3 below: Where k is the number of features entering the model, and k is preferably 6; α is the intercept term, the value range is (-0.11402, -0.35754), and the optimal value is -0.2357787; β1 is the coefficient corresponding to the maximum number of consecutive months of increase in asset size in the past three months, with a value range of (-0.19897, -0.31636) and an optimal value of -0.2576679; β2 is the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months, with a value range of (-0.03843, -0.28001), and the optimal value is -0.1592187; β3 is the coefficient corresponding to the proportion of months with a credit limit utilization rate of >70% in the past three months, with a value range of (-0.71885, -0.91682) and an optimal value of -0.8178348; β4 is the coefficient corresponding to the minimum credit card repayment rate in the past 6 months, with a value range of (-0.18116, -0.49436) and an optimal value of -0.3377601; β5 is the coefficient corresponding to the number of months from now when the credit card interest rate in the past three months is greater than 0, with a value range of (-0.22644, -0.47791), and the optimal value is -0.352176; β6 is the percentage coefficient of the total amount of cash withdrawals from credit card consumption in the current month (including installments) / the total amount of cash withdrawals from consumption in the previous month (including installments). The value range is (-0.05659, -0.38296), and the optimal value is -0.2197794; x1 is the natural logarithm-transformed value of the maximum number of consecutive months of increase in asset size over the past three months generated in the feature conversion step; x2 is the original value of the percentage of overdue interest on mortgage loans in the past six months to all repayment amounts in the past six months generated by the feature conversion step; x3 is the cube root conversion value of the percentage of months with a credit limit utilization rate > 70% in the past three months generated by the feature conversion step; x4 is the dummy variable conversion value of the minimum credit card repayment rate in the past 6 months generated in the feature conversion step; x5 is the natural logarithm transformation value of the number of months from now when the credit card interest rate in the past three months is >0, generated in the feature conversion step; x6 is a feature conversion step that generates a WOE conversion value of the percentage of the total amount of cash withdrawals from credit card consumption in the current month (including installments) / the total amount of cash withdrawals from consumption in the previous month (including installments).

24. The method according to claim 23, wherein: After calculating the credit default probability, the step of calculating the credit score of the sample to be predicted is to use the following formula to calculate and generate a credit score for characterizing the borrower: Among them, P is the borrower's default probability (P) generated in the credit default probability calculation module, A is 54.2458; B is 115.4156, and the round function rounds the calculated score to the integer value; finally, the score greater than 1000 is set to 1000 points, and the score less than 0 is set to 0 points.

25. The method according to any one of claims 1 to 9, wherein: The retail credit forecast data is selected from: the deposit balance at the current point in time, the average balance of investment and financial management accounts in the past 12 months, the proportion of the largest number of consecutive months of asset size increase in the past three months, the current remaining available credit card limit, the proportion of months with credit card repayment rate < 90% in the past 12 months, the largest number of consecutive months of increase in credit card consumption balance (including consumption installments) in the past three months, and one, two, three, four, five, six or seven of the number of months of credit card revolving and installment use in the past six months.

26. The method according to claim 25, wherein: The steps to calculate the probability of credit default include: The current deposit balance, the average balance of investment and financial management accounts in the past 12 months, the proportion of the largest consecutive increase in asset size in the past three months, the current remaining available credit card limit, the proportion of months with a credit card repayment rate < 90% in the past 12 months, the largest consecutive increase in credit card consumption balance (including consumption installments) in the past three months, and the number of months of credit card cycle and installment use in the past six months are converted into features. Preferably, The current deposit balance is converted using a continuous method; the average balance of investment and financial management accounts in the past 12 months is converted using a continuous method; the proportion of the largest consecutive increase in asset size in the past three months is converted using a continuous method; the current remaining available credit card amount is converted using a continuous method; the proportion of the number of months with a credit card repayment rate <90% in the past 12 months is converted using a continuous method; the maximum number of consecutive increase in credit card consumption balances (including consumption installments) in the past three months is converted using a dummy variable method; the number of months of credit card cycle and installment use in the past six months is converted using a continuous method; More preferably, The current deposit balance is converted using a continuous method to a calculation method using the natural logarithm of the current deposit balance; the average balance of the investment and financial management account in the past 12 months is converted using a continuous method to a calculation method using the natural logarithm of the average balance of the investment and financial management account in the past 12 months; the proportion of the largest consecutive increase in the asset size value in the past 3 months is converted using a continuous method to a calculation method using the original value; the current remaining available credit limit of the credit card is converted using a continuous method to a calculation method using the square root of the current remaining available credit limit; the proportion of the number of months with a credit card repayment rate of <90% in the past 12 months is converted using a continuous method to a calculation method using the original value of the proportion of the number of months with a credit card repayment rate of <90% in the past 12 months; the number of months of credit card revolving and installment use in the past 6 months is converted using a continuous method to a calculation method using the cube root of the number of months of credit card revolving and installment use in the past 6 months.

27. The method according to claim 25, wherein: The converted values ​​of the seven features, namely, the current deposit balance, the average balance of investment and wealth management accounts in the past 12 months, the proportion of months with the largest consecutive increase in asset size in the past three months, the current remaining available credit card limit, the proportion of months with credit card repayment rate < 90% in the past 12 months, the maximum number of consecutive months with credit card consumption balance (including consumption installments) in the past three months, and the number of months of credit card revolving and installment use in the past six months, are substituted into the sub-model constructed by logistic regression based on the sample retail credit prediction data and credit default probability to calculate the default probability of the sample to be predicted.

28. The method according to claim 27, wherein: The sub-model is shown in Formula 4 below: Where k is the number of features entering the model, and k is preferably 7; α is the intercept term, the value range is (-0.12673,-0.19974), and the optimal value is -0.1632328; β1 is the coefficient corresponding to the deposit balance at the current time point, with a value range of (-0.44362, -0.53469), and the optimal value is -0.4891512; β2 is the coefficient corresponding to the average balance of investment and financial management accounts in the past 12 months, with a value range of (-0.181, -0.52118) and an optimal value of -0.3510896; β3 is the coefficient corresponding to the proportion of the largest consecutive increase in asset size in the past three months, with a value range of (0.77324, 0.69876) and an optimal value of 0.736; β4 is the coefficient corresponding to the current remaining available credit limit of the credit card, with a value range of (-0.16337, -0.27663), and the optimal value is -0.22; β5 is the coefficient corresponding to the proportion of months with credit card repayment rate < 90% in the past 12 months, with a value range of (0.1674, 0.0106) and an optimal value of 0.089; β6 is the coefficient corresponding to the maximum number of consecutive months of increase in credit card consumption balance (including consumption installments) in the past three months, with a value range of (-0.57383, -0.78447) and an optimal value of -0.6791512; β7 is the coefficient corresponding to the number of months of credit card cycle and installment use in the past 6 months, with a value range of (-0.01967, -0.06671) and an optimal value of -0.0431908; x1 is the natural logarithm transformation value of the deposit balance at the current time point generated by the feature transformation step; x2 is the natural logarithm transformation value of the average balance of investment and financial management accounts in the past 12 months generated in the feature conversion step; x3 is the original value of the proportion of the largest consecutive increase in asset size in the past three months generated by the feature conversion step; x4 is the square root conversion value of the current remaining available credit limit of the credit card generated in the feature conversion step; x5 is the original value of the percentage of months with a credit card repayment rate < 90% in the past 12 months generated by the feature conversion step; x6 is the dummy variable conversion value of the maximum number of consecutive months of credit card consumption balance (including consumption installments) in the past three months generated by the feature conversion step; x7 is the cube root conversion value of the number of months of credit card recurrence and installment use in the past 6 months generated in the feature conversion step.

29. The method according to claim 28, wherein: After calculating the credit default probability, the step of calculating the credit score of the sample to be predicted is to use the following formula to calculate and generate a credit score for characterizing the borrower: Among them, P is the borrower's default probability (P) generated in the credit default probability calculation module, A is 54.2458; B is 115.4156, and the round function rounds the calculated score to the integer value; finally, the score greater than 1000 is set to 1000 points, and the score less than 0 is set to 0 points.

30. The method according to any one of claims 1 to 9, wherein: The retail credit forecast data is selected from: one, two, three or four of the following: the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, the maximum salary in the past 12 months, the maximum balance of the investment and financial management account in the past 6 months, and the minimum balance of the deposit account at a certain point in time in the past 6 months.

31. The method according to claim 30, wherein: The steps to calculate the probability of credit default include: The average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, the maximum salary in the past 12 months, the maximum balance of the investment and financial management account in the past 6 months, and the minimum balance of the deposit account at a certain point in time in the past 6 months are converted into features. Preferably, The average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months are converted in a continuous manner; the maximum salary in the past 12 months are converted in a continuous manner; the maximum balance of the investment and financial management account in the past 6 months is converted in a dummy variable manner; the minimum balance of the deposit account at a certain point in time in the past 6 months is converted in a continuous manner; Further pre-selected, The average monthly asset size of the past 6 months / the average monthly asset size of the past 12 months is converted into the average monthly asset size of the past 6 months / the average monthly asset size of the past 12 months using the original value calculation method; the maximum salary of the past 12 months is converted into the maximum salary of the past 12 months using the natural logarithm calculation method; the minimum balance of the deposit account at a certain point in time in the past 6 months is converted into the minimum balance of the deposit account at a certain point in time in the past 6 months using the natural logarithm calculation method.

32. The method of claim 30, wherein: The converted values ​​of the four features, namely, the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, the maximum salary in the past 12 months, the maximum balance of the investment and financial management account in the past 6 months, and the minimum balance of the deposit account at a certain point in time in the past 6 months, are substituted into the sub-model constructed by logistic regression based on the sample retail credit prediction data and credit default probability to calculate the default probability of the sample to be predicted.

33. The method of claim 32, wherein: The sub-model is shown in Formula 5 below: Where k is the number of features entering the model, preferably k is 4; α is the intercept term, the value range is (-2.04845, -2.31097), and the optimal value is -2.1797076; β1 is the coefficient of the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months, with a value range of (-0.20978, -0.34796) and an optimal value of -0.2788716; β2 is the coefficient corresponding to the maximum salary in the past 12 months, with a value range of (-0.05551, -0.39342) and an optimal value of -0.2244648; β3 is the coefficient corresponding to the maximum balance of the investment and financial management account in the past 6 months, with a value range of (-0.1207, -0.12475) and an optimal value of -0.122724; β4 is the coefficient corresponding to the minimum balance of the deposit account at a certain point in time in the past 6 months. The value range is (-0.68315, -0.93368), and the optimal value is -0.808416. x1 is the original value of the average monthly asset size in the past 6 months / the average monthly asset size in the past 12 months generated by the feature conversion step; x2 is the natural logarithm-transformed value of the maximum salary in the past 12 months generated in the feature transformation step; x3 is the dummy variable conversion value of the maximum balance of the investment and financial management account in the past 6 months generated in the feature conversion step; x4 is the natural logarithm-transformed value of the minimum balance of the deposit account at a certain point in time in the past six months generated in the feature conversion step.

34. The method of claim 33, wherein: After calculating the credit default probability, the step of calculating the credit score of the sample to be predicted is to use the following formula to calculate and generate a credit score for characterizing the borrower: Among them, P is the borrower's default probability (P) generated in the credit default probability calculation module, A is 54.2458; B is 115.4156, and the round function rounds the calculated score to the integer value; finally, the score greater than 1000 is set to 1000 points, and the score less than 0 is set to 0 points.

35. A device for calculating retail credit risk, comprising: A data collection module, which is used to obtain retail credit prediction data of samples to be predicted; A module for classifying samples to be predicted, which is used to classify the samples to be predicted based on a decision tree method to determine a sub-model for calculating the probability of credit default; The credit default probability calculation module is used to substitute the retail credit prediction data into the credit default probability sub-model to calculate the credit default probability of the sample to be predicted.

36. The device according to claim 35, wherein The device executes the steps of the method for calculating retail credit risk according to any one of claims 1 to 34.

37. A system for calculating retail credit risk, wherein: The system for calculating retail credit risk comprises: a memory, a processor, and a program for the method for calculating retail credit risk stored in the memory and executable on the processor. When the program for calculating retail credit risk is executed by the processor, the steps of the method for calculating retail credit risk as described in any one of claims 1 to 34 are implemented.

38. A computer storage medium, wherein: The computer storage medium stores a program for calculating the retail credit risk method, and when the program for calculating the retail credit risk method is executed by the processor, the steps of the method for calculating the retail credit risk according to any one of claims 1 to 34 are implemented.

39. A method for constructing a retail credit risk prediction model for credit card business, comprising: A data collection step, which obtains the original retail credit prediction data of the sample used to build the model; A data derivation step, which processes derived retail credit forecast data based on the original retail credit forecast data; The feature initial screening step is to perform initial screening on all categories, that is, all features, including the original retail credit forecast data and the derived retail credit forecast data, to obtain features after initial screening; The initial screening data conversion step is to judge the conversion method of the features after the initial screening to confirm whether to use one of the WOE conversion method, the dummy feature conversion method and the continuous conversion method for feature conversion, and use the best method to convert each feature after the initial screening; The feature fine screening step is to perform in-depth screening on the initially screened features after feature conversion to obtain finely screened features; The credit default probability modeling step is to select the logistic regression method to build the model based on the relationship between the carefully screened features and the probability of credit default, and confirm the method used to calculate the credit default probability; The data samples obtained in the data collection step are samples of customers who have used credit card services.

40. The method of claim 39, wherein: In the data collection step, the original retail credit prediction data of the samples obtained for building the model include: Credit card basic data, which is based on all available data during the sample user's credit card creation and usage process. Basic data on personal loans, which is all available data based on the loan application and usage behavior of sample users. Basic customer information data, which is based on the attributes of the sample users themselves but is not directly related to their behavior in financial institutions. Basic data on personal financial assets, which includes all other financial assets and financial transaction data of sample users in financial institutions that are not related to credit cards and loans.

41. The method of claim 39, wherein: In the data derivation step, the derived retail credit prediction data processed based on the original retail credit prediction data refers to data obtained by processing the collected original retail credit prediction data based on the time dimension, space dimension, frequency dimension, and statistical information dimension; Preferably, the derived retail credit forecast data includes but is not limited to: Derived retail credit forecast data obtained by processing based on sample relationship length, Derivative retail credit forecast data obtained by processing time interval variables, Derivative retail credit forecast data obtained by processing based on the frequency of sample behavior, Derivative retail credit forecast data obtained by processing the current time point of the sample, Derivative retail credit forecast data obtained by processing based on sample continuous behavior, Derived retail credit forecast data obtained by processing sample data based on statistical information dimensions.

42. The method according to any one of claims 39 to 41, wherein: The feature initial screening step includes the following steps: The first preliminary screening step is to screen the features based on the missing data of each feature of the sample used to build the model. The second initial screening step is to screen the features based on the fact that a single value of a feature sample is too high. The third preliminary screening step is to calculate the information IV value of each feature to perform preliminary screening of the features; The order of the first preliminary screening step, the second preliminary screening step and the third preliminary screening step can be any order. The fourth preliminary screening step uses a stepwise discrimination algorithm to perform preliminary screening of features after the first to third preliminary screenings; The fifth preliminary screening step is to conduct preliminary screening of the features after the fourth preliminary screening step based on the consistency between the risk characteristics of each feature itself and the actual real results of the samples used for model construction.

43. The method according to any one of claims 39 to 42, wherein: Also includes: The sample selection step is used to screen all users before the data collection step and obtain samples using credit card services for model construction. Preferably, the sample selection step includes classifying all sample users based on a decision tree, and the classification basis includes but is not limited to: Whether a user is a customer who has applied for credit registration at a financial institution; Whether the account age of a certain user is greater than or equal to G1; Whether a user is a customer with overdue credit loans; Whether the current credit limit usage rate of a certain user is less than a%.

44. The method according to any one of claims 39 to 43, wherein: In the initial screening data conversion step, the conversion method of the features after the initial screening is determined based on the concentration and data type of the features after the initial screening.

45. The method of claim 44, wherein: The initial screening data conversion step includes the following steps based on the judgment of concentration and data type: Classify the data type of each feature into character variables and numeric variables. For character variables, dummy feature conversion is used to perform initial screening data conversion. The process of further classifying numerical variables includes the following sub-steps: If the value of the numeric variable is less than n, the WOE conversion method is used to convert the initial screening data. If the value of the numerical variable is more than n, further judge if the value of the continuous variable is large and the concentration of a single value is greater than m%, then the WOE conversion method is used; if the concentration of a single value is less than or equal to m%, then the continuous conversion method is used. Preferably, n and m are both positive integers, wherein n=5-10, and m=90-99.

46. ​​The method of claim 45, wherein: Also includes: For the features that are confirmed to adopt the continuous conversion method, the optimal conversion method is selected based on the correlation between the feature and credit default under different continuous conversion methods to perform the continuous feature conversion of the feature. Preferably, the continuous feature conversion is performed by directly selecting the original value, calculating the square of the original data, calculating the square root of the original data, calculating the cube root of the original data, or calculating the natural logarithm of the original data.

47. The method according to any one of claims 39 to 46, wherein: The feature fine screening steps include: The first fine screening step is to screen the features based on the stepwise regression algorithm, the F test and the T test to determine the significance of the features. The second fine screening step is to calculate the variance inflation factor based on each feature and eliminate the features with higher variance inflation factors to screen the features. The third fine screening step is to analyze whether the characteristic coefficients conform to the trend of the prediction results for credit default based on logistic regression of the characteristics after the first fine screening step and the second fine screening step, so as to further perform feature screening.

48. The method according to any one of claims 39 to 47, wherein: The credit default probability modeling step substitutes the features selected in the feature fine screening step into the Sigmoid function to perform logistic regression to calculate the credit default probability model.

49. A device for constructing a retail credit risk prediction model for credit card business, wherein: The device comprises: A data collection module, which is used to obtain raw retail credit prediction data of samples used to build a model; A data derivation module, which is used to process derived retail credit forecast data based on the original retail credit forecast data; A feature initial screening module, which is used to perform initial screening on all categories, that is, all features, including the original retail credit forecast data and the derived retail credit forecast data, to obtain features after initial screening; The initial screening data conversion module is used to determine the conversion method of the features after the initial screening to confirm whether to use one of the WOE conversion method, the dummy feature conversion method and the continuous conversion method for feature conversion, and to use the best method for feature conversion for each feature after the initial screening; A feature fine screening module is used to perform in-depth screening on the initially screened features after feature conversion to obtain finely screened features; The credit default probability modeling module is used to select the logistic regression method to build a model based on the probability relationship between the finely screened features and credit default, and confirm the method used to calculate the credit default probability. The data samples obtained by the data collection module are samples of customers who have used credit card services.

50. The apparatus of claim 49, wherein: The device executes the steps of the method for constructing a retail credit risk prediction model as described in any one of claims 39 to 48.

51. A system for constructing a retail credit risk prediction model for credit card business, wherein: The system includes: a memory, a processor, and a program for constructing a retail credit risk prediction model stored in the memory and executable on the processor. When the program for constructing a retail credit risk prediction model is executed by the processor, the steps of constructing a retail credit risk prediction model as described in any one of claims 39 to 48 are implemented.

Images