Method and device for predicting credit line of user, and medium

Abstract

The invention provides a user credit line prediction method and device and a medium, and the method comprises the steps: obtaining first credit data and second credit data of a user based on a loan request of the user, and obtaining a credit index according to the first credit data and the second credit data, obtaining whether the user has credit granting qualification according to the credit granting index; if the user does not have the credit granting qualification, refusing the credit granting request, and if the user has the credit granting qualification, obtaining the income and expenditure flow data of the user and the default probability; obtaining an initial credit line prediction value of the user by adopting a pre-constructed line prediction model according to the income and expenditure flow data, and obtaining a final credit line prediction value based on the initial credit line prediction value and the default probability in combination with the capital yield of the lending platform, the technical scheme provided by the invention comprehensively considers the credit risk of the user and the capital income of the lending platform.

Description

Technical Field

[0001] This document relates to the technical field of credit limits, and particularly to a method, device, and medium for predicting user credit limits. Background Art

[0002] Customer credit granting has always been an important part of the consumer finance field. The credit limit is also the core focus of research for various lending platforms and consumer finance companies. If the credit limit is too high, the customer churn rate will decrease, but the risk of bad debts will increase; if the credit limit is too low, although the risk of bad debts is reduced, the customer churn rate will increase and the business volume will decline. For example, if a customer is not satisfied with the credit card limit given by a lending platform, they will not choose to register and use the credit card. In traditional methods for obtaining credit limits, usually, information such as the customer's asset holdings and default risk is obtained based on the customer's credit investigation analysis, and the corresponding credit limit is further determined based on information such as asset holdings and default risk and informed to the customer. If the customer is not satisfied with the credit limit, they will abandon using the credit limit. However, since the credit granting process in existing methods for obtaining credit limits is relatively single, it is impossible to accurately obtain a credit limit that matches the customer based on more comprehensive customer information. Therefore, there is a problem in existing methods for obtaining credit limits that the credit limit cannot be accurately obtained.

[0003] In the operation process of commercial lending platforms, the credit granting business guarantees the profit source of the lending platforms. Under the condition of a certain loan interest rate, it not only determines the amount of income of the lending platforms but also controls risks to a certain extent; for enterprises applying for credit granting business, the credit granting business is the source of funds to guarantee enterprise production and operation and expand scale. Therefore, the research on the calculation of credit limits is crucial for the development of commercial lending platforms. Summary of the Invention

[0004] Embodiments of the present invention provide a method, device, and medium for predicting user credit limits, aiming to solve the above problems.

[0005] An embodiment of the present invention provides a method for predicting user credit limits, including:

[0006] S1. Obtain the first credit data and the second credit data of the user based on the user's loan request, obtain a credit index according to the first credit data and the second credit data, and determine whether the user is eligible for credit according to the credit index;

[0007] S2. If the user is not eligible for credit, reject the credit request; if the user is eligible for credit, obtain the user's income and expenditure flow data and the default probability;

[0008] S3. Obtain the predicted value of the initial credit limit of the user using the pre-constructed limit prediction model based on the income and expenditure transaction data, and obtain the predicted value of the final credit limit by combining the predicted value of the initial credit limit and the default probability with the capital return rate of the lending platform.

[0009] One or more embodiments of this specification provide an electronic device, including:

[0010] A processor; and,

[0011] A memory arranged to store computer-executable instructions, and when the computer-executable instructions are executed, the processor is caused to execute the steps of the method for predicting the credit limit of the user as described above.

[0012] One or more embodiments of this specification provide a storage medium for storing computer-executable instructions, and when the computer-executable instructions are executed, the steps of the method for predicting the credit limit of the user as described above are implemented.

[0013] By adopting the embodiments of the present invention, a decision on whether to have the qualification for credit is initially made by considering the person to be credited and the comprehensive credit information related to the person to be credited. If the person does not have the qualification for credit, there is no need to continue collecting the income and expenditure transaction data of the person to be credited, which can more comprehensively consider the credit level of the person to be credited. When giving the predicted value of the final credit limit, the risk-bearing capacity of the credit-granting institution is also fully considered, and the predicted value of the final credit limit is obtained by combining the capital return rate of the lending platform, thereby improving the accuracy of the predicted value of the credit limit of this application. Description of the Drawings

[0014] In order to more clearly illustrate the technical solutions in one or more embodiments of this specification or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in this specification. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0015] Figure 1 It is a flowchart of the method for predicting the credit limit of the user in the embodiments of the present invention. Detailed Embodiments

[0016] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will clearly and completely describe the technical solutions in one or more embodiments of this specification in conjunction with the accompanying drawings in one or more embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this document.

[0017] Method Embodiment

[0018] According to an embodiment of the present invention, a method for predicting a user's credit limit is provided. Figure 1 This is a flowchart of the method for predicting a user's credit limit according to an embodiment of the present invention. According to Figure 1 as shown, the method for predicting a user's credit limit according to an embodiment of the present invention specifically includes:

[0019] S1. Obtain the first credit data and the second credit data of the user based on the user's loan request, obtain a credit index according to the first credit data and the second credit data, and determine whether the user is eligible for credit according to the credit index;

[0020] Types of user loan requests: Users may apply for different types of loans due to different needs, such as housing loans, car loans, personal consumption loans, business loans, etc. Each loan type has different risk assessment criteria and requirements.

[0021] The first credit data refers to information used to evaluate the credit status of an individual or an enterprise. The "first credit data" mentioned here may refer to basic information directly related to the user's loan request, such as the user's identity information, income status, job stability, past loan records (if any), credit card usage, repayment records, etc. This data usually comes from the internal databases of credit reporting agencies or financial institutions.

[0022] Score according to the first preset conditions using the first credit data:

[0023] Preset first preset conditions: These conditions refer to a series of scoring criteria and weights set in advance according to the loan type and risk assessment requirements. These conditions may include multiple dimensions such as income stability, debt ratio, credit history length, number of overdue payments, number of inquiries, etc.

[0024] The scoring process is to match the user's credit data with the preset conditions and calculate the corresponding scores according to the weight of each condition and the actual situation of the user. This process usually involves complex algorithms and models to ensure the accuracy and fairness of the scoring.

[0025] Obtain the first credit score:

[0026] First credit score: After the above scoring process, the obtained score is the user's "first credit score". This score reflects the user's credit status and risk level under the current loan request type.

[0027] Application of credit score: Credit score plays a crucial role in the loan approval process. Financial institutions usually determine whether to approve a loan application, loan amount, interest rate and other conditions based on the credit score. Users with higher credit scores are usually more likely to obtain loan approval and enjoy more favorable loan conditions.

[0028] The specific steps of S1 include:

[0029] S11. Score according to the first credit investigation data using a preset first preset condition to obtain the first credit score. If the first credit score is less than the first preset credit value, reject the credit request;

[0030] The specific steps of S11 include:

[0031] Obtain the user's credit report data from the People's Bank of China and related persons. The user's credit report data from the People's Bank of China includes basic data and classified data. The basic data includes age, education level, educational background, marital status and work information;

[0032] According to the classified data, classify the classified data corresponding to each basic data into a first-level classification, a second-level classification corresponding to the first-level classification, and a feature field corresponding to the second-level classification;

[0033] Among them, the first-level classification includes account type, business type and account status. The second-level classification includes non-revolving loan accounts, revolving loan accounts and credit card accounts under the account type, housing commercial loans, housing provident fund loans and consumer loans under the business type, overdue accounts, settled and unsettled under the account status. The feature fields include: original variables and derived variables. Among them, the original variables include: variables mentioned in the user's credit report data from the People's Bank of China, including monthly consumption records of each credit card, newly opened accounts, auto loan information and mortgage loan information. The derived variables are new variables calculated based on the original variables, including the total credit card consumption amount and the total number of accounts;

[0034] After processing the outlier and missing value information of the classified credit report data from the People's Bank of China, score the user's credit report data from the People's Bank of China to obtain the first credit score.

[0035] S12. If the first credit score is greater than the first preset credit value and less than the second preset credit value, obtain the second credit investigation data, score according to the second credit investigation data using the preset second preset condition to obtain the second credit score, obtain the sum of the first credit score and the second credit score, and determine whether the credit granting qualification is available.

[0036] The second credit investigation data is the first credit investigation data of the related person.

[0037] The obtaining the sum of the first credit score and the second credit score and determining whether the credit granting qualification is available specifically includes:

[0038] Compare the sum of the first credit score and the second credit score with the first preset credit value. If the sum of the first credit score and the second credit score is greater than the first preset credit value, obtain the credit granting qualification; otherwise, reject the credit granting request.

[0039] S13. If the first credit score is greater than the second preset credit value, obtain the credit granting qualification.

[0040] If the credit investigation report result of the borrower does not meet the first preset credit value, such as the loan is overdue more than A times, directly reject the credit granting request. If the credit investigation report of the borrower shows that the number of loan overdue times is small, the credit situation of the related person with the borrower, such as spouse, parents, etc., can be further referred to. If the credit investigation report of the related person with the borrower is good, the credit granting qualification can be given according to the specific situation.

[0041] A specific first credit investigation data scoring model of the embodiment of the present invention is as follows:

[0042] 1. First, select indicators. The embodiment of the present invention selects basic information type, credit history type, and income and liability type data as the evaluation indicators for the first credit investigation data scoring:

[0043] The basic information type includes: age, gender, and occupation. The repayment ability and credit risk vary in different age groups. For example, young people may have unstable incomes but great potential, while the elderly may have stable incomes but may face risks such as health affecting the repayment ability. Some studies show that there may be certain differences in credit behaviors between different genders, but the differences are relatively small. Stable occupations (such as civil servants, teachers, etc.) usually have lower credit risks, while self-employed or practitioners in high-risk industries may have higher credit risks.

[0044] The credit history type includes: credit record duration, number of overdue times, and amount of arrears. The longer the credit record, the more it can reflect the long-term credit behavior. Generally, people with a long and good credit record have lower credit risks; the more the number of overdue times, the higher the credit risk; amount of arrears: the higher the proportion of the amount of arrears to the income, the greater the repayment pressure and the higher the credit risk.

[0045] Income and liability data include: Annual income: A higher annual income usually means stronger repayment ability. Debt-to-income ratio: Reflects the relationship between an individual's debt level and income. The higher the ratio, the greater the repayment pressure.

[0046] 2. Then perform weight setting. In the embodiments of the present invention, the weight setting can use methods such as expert scoring method and analytic hierarchy process to determine the weights of each index. The following is an example weight:

[0047] Table 1 Weight Setting of the First Credit Data Scoring Index

[0048]

[0049]

[0050] 3. Finally, perform quantitative scoring on each index. For example:

[0051] Age: 70 points for 20 - 30 years old, 80 points for 31 - 40 years old, 85 points for 41 - 50 years old, 80 points for 51 - 60 years old, and 70 points for others.

[0052] Number of overdue times: 100 points for 0 times, 80 points for 1 - 2 times, 60 points for 3 - 5 times, and 30 points for more than 5 times.

[0053] Multiply the quantitative scores of each index by their weights, and then sum them up to obtain the first credit score.

[0054] The second credit data scoring model of the embodiments of the present invention is as follows:

[0055] First, perform index selection. In the embodiments of the present invention, credit history data and asset data are selected as the evaluation indexes for the second credit data scoring:

[0056] Credit history data includes: Credit record duration, number of overdue times, and credit card usage; The credit record duration is similar to that in the first credit data, reflecting long-term credit behavior; The number of overdue times reflects credit default situations. Credit card usage is used to obtain credit card limit utilization rate, whether the payment is made on time, etc.

[0057] Asset data includes: Real estate situation and vehicle situation. Real estate situation: Owning real estate usually means stronger economic strength and stability. Vehicle situation: Vehicle value and property rights situation can also reflect an individual's economic status.

[0058] 2. Then perform weight setting. In the embodiments of the present invention, the weight setting can use methods such as expert scoring method and analytic hierarchy process to determine the weights of each index. The following is an example weight:

[0059] Table 2 Weight Setting of the Second Credit Investigation Data Scoring Index

[0060]

[0061]

[0062] 3. Finally, perform the scoring calculation. Similarly, perform a quantitative score for each index. For example:

[0063] Length of credit record: 90 points for over 5 years, 80 points for 3 - 5 years, 70 points for 1 - 3 years, and 60 points for less than 1 year.

[0064] Housing situation: 100 points for own property, 60 points for renting.

[0065] Multiply the quantitative scores of each index by their weights, and then sum them up to obtain the second credit score.

[0066] The second credit investigation data scoring model in the embodiments of the present invention can also adopt the same method as the first credit investigation data scoring model.

[0067] S2. If the user does not have the credit granting qualification, reject the credit granting request. If the user has the credit granting qualification, obtain the user's income and expenditure flow data and the default probability;

[0068] After the preliminary determination in S1 above, when the person to be credit - granted has the preliminary credit - granting qualification, further obtain the income and expenditure flow data of the person to be credit - granted, and further determine the default probability of the person to be credit - granted by combining the income and expenditure flow data with the credit investigation data.

[0069] S3. According to the income and expenditure flow data, use the pre - constructed quota prediction model to obtain the initial credit quota prediction value of the user, and based on the initial credit quota prediction value and the default probability, combine with the capital return rate of the lending platform to obtain the final credit quota prediction value.

[0070] The pre - constructed quota prediction model is constructed through the following steps:

[0071] Obtain the modeling variables, specifically including: perform feature derivation according to the income and expenditure flow data and loan request data of historical credit - granted users to obtain feature - derived variables, and the feature - derived variables include: average monthly deposit balance of historical credit - granted users, tax expenditure amount of historical credit - granted users, operating activity income amount of historical credit - granted users, and transaction expenditure amount of historical credit - granted users; determine the loan data information of the historical credit - granted users according to the income and expenditure flow data of the historical credit - granted users; perform the first screening on the feature - derived variables according to the feature missing rate of the feature - derived variables and the correlation coefficient between the feature - derived variables and the loan data information; perform the second screening on the feature - derived variables obtained after the first screening according to the business meaning to obtain the modeling variables;

[0072] Obtain a quota prediction model, specifically including: screening a preset number of historical credit-granted users according to preset screening conditions to obtain a training set of historical credit-granted users and a test set of historical credit-granted users, training the modeling variables of the training set and the test set through a linear regression model established by the Lasso regression method to obtain a quota model;

[0073] The process of screening a preset number of historical credit-granted users according to preset screening conditions to obtain a training set of historical credit-granted users and a test set of historical credit-granted users specifically includes: screening historical credit-granted users according to historical credit-granted user screening conditions to obtain a sample of historical credit-granted users. The historical credit-granted user screening conditions include: the loan date of historical credit-granted users, the scale of historical credit-granted users, the loan amount of historical credit-granted users, the loan data information of historical credit-granted users, outliers of loan data information, and bad credit; determining the cross-time validation set of the historical credit-granted users and the combined set of the training set and the test set according to a preset time point; determining the training set of historical credit-granted users and the test set of historical credit-granted users in the combined set according to a preset ratio.

[0074] The predicted credit quota value of the user includes: credit quota, repayment method, repayment period, and loan interest rate.

[0075] The process of obtaining the final predicted credit quota value based on the initial predicted credit quota value and the default probability in combination with the capital return rate of the lending platform specifically includes:

[0076] Calculate the capital return rate of the loan according to the loan policy, interest rate setting, and operating cost of the lending platform;

[0077] Take the initial predicted credit quota value, default probability, and capital return rate as input variables, input them into a comprehensive evaluation model, and set different weights according to the lending platform's preference for risk and return to obtain the final predicted credit quota value.

[0078] In a specific implementation of the embodiment of the present invention, it is considered to use the RAROC model to obtain the capital return rate of the lending platform. The RAROC model includes three different dimensions of indicators: interest income, capital, and risk. In the formula of the RAROC model, the numerator is the risk-adjusted net income measured by the commercial lending platform, usually expressed as interest income minus the cost of funds, operating expenses, and expected losses. The denominator is the economic capital, that is, the unexpected loss, which is the corresponding capital that the lending platform needs to extract when facing unpredictable risks. The RAROC model reflects the efficiency of the commercial lending platform in using its economic capital.

[0079] Judging from the actual situation of most domestic lending platforms, since the special reserve that each commercial lending platform should set aside is not deducted from the total profit. Therefore, the risk-adjusted return on assets can also be obtained in the following ways. In this paper, the average value of RAROC is calculated through the formula::

[0080]

[0081] RAROC is mainly calculated based on four parameters: the cost of funds, operating costs, expected losses, and economic capital of the lending platform::

[0082] 1. Cost of funds::

[0083] The cost of funds of a commercial lending platform consists of the average cost of funds and the marginal cost of funds. The concept of the marginal cost of funds is

[0084] The interest expenditure and expense expenditure required to obtain one unit of investable funds. The average cost of funds depends on the source of funds and does not take into account subsequent changes in interest rates and fees. If you want to measure the operating conditions of a commercial lending platform over the years, the average cost of funds can be used for evaluation. The funds used by commercial lending platforms to support credit business come from two aspects: one is to absorb deposits, and the other is to borrow from the central lending platform and inter-bank lending business among lending platforms. In the process of financial management, the deposits absorbed and borrowed funds need to be recorded as interest-bearing liabilities in the balance sheet of commercial lending platforms. Therefore, the cost of funds rate needs to be calculated by weighted averaging the costs of customer deposits and borrowed funds. This paper mainly calculates the cost of funds based on the expenses incurred for increasing investment funds in a single credit business. The calculation formula is: Cost of funds = Total credit amount * Cost of funds rate = L * r, where L is the total amount of loans and advances issued by the commercial lending platform on the balance sheet, and r is the cost of funds rate.

[0085] 2. Operating costs::

[0086] The definition of operating costs refers to all costs that a company should bear during its operation, including taxes, costs, and period expenses incurred during the sales process. Since commercial lending platforms belong to financial institutions and their profit model is mainly based on the interest rate spread between deposits and loans. Therefore, business management fees are the source of operating costs for commercial lending platforms. The calculation formula for operating costs is: Operating costs = Total loan amount * Operating cost rate = L * c, where c is the operating expense rate.

[0087] 3. Economic capital::

[0088] Economic capital is the same as risk capital. Its concept refers to the funds invested by shareholders to bear operating risks or purchase external returns. It is a type of capital evaluated by the internal management of a commercial lending platform and is used to allocate to assets or businesses to reduce the impact of risky investments. Economic capital is the capital required to bear unexpected losses and maintain the normal operation of a commercial lending platform. From the perspective of a commercial lending platform, economic capital is the minimum capital to ensure the normal operation of the lending platform and the last resort to prevent the capital chain from breaking.

[0089] The calculation of economic capital is as follows: Economic capital = Unexpected loss of credit risk + Unexpected loss of market risk + Unexpected loss of operational risk.

[0090] 4. Expected loss:

[0091] Expected loss represents the risk cost. Its concept is the average loss expected to be suffered by a commercial lending platform during a certain period. As an indicator reflecting credit risk, usually the actual loss fluctuates around the average level. Therefore, the average loss value can be considered relatively certain during the measurement process. In the total cost of the lending platform, the expected loss exists in the form of reserves.

[0092] The calculation of expected loss is as follows: Expected loss = Probability of default * Loan balance at default * Loss given default = PD * L * LGD, where PD represents the probability of default, L represents the loan balance at default, and LGD represents the loss given default. The probability of default (PD) refers to the possibility that a debtor is unable to repay the principal and interest of the loan from the commercial lending platform within the borrowing contract period or fulfill the relevant obligations within the borrowing contract in the future.

[0093] Loss given default (LGD) represents the severity of the loss, which is the ratio of the asset loss caused to the creditor when the debtor fails to fulfill the obligations according to the borrowing contract. For corporate credit granting business, it is the ratio of the loss amount of the commercial lending platform after the enterprise defaults to the total risk exposure of the commercial lending platform before the enterprise defaults. The loss given default rate is an important parameter in the international lending platform supervision system. The calculation formula for the risk-adjusted return on capital is as follows:

[0094]

[0095] Derive the formula for calculating the credit limit based on the RAROC model:

[0096]

[0097] In this paper, a quantitative analysis method is used to establish a model for predicting the probability of default behavior of loan - applying enterprises, quantify the default risks borne by lending platforms, and comprehensively and scientifically measure risks, making the risks statistically significant to a certain extent, so as to improve the credit management and risk control levels of commercial lending platforms.

[0098] By adopting the embodiments of the present invention, the following beneficial effects are achieved:

[0099] By adopting the embodiments of the present invention, a decision on whether to have the qualification for credit granting is initially made by considering the comprehensive credit information of the person to be credit - granted and the associated persons related to the person to be credit - granted. If the qualification for credit granting is not met, there is no need to continue collecting the income and expenditure flow data of the person to be credit - granted. The credit level of the person to be credit - granted can be considered more comprehensively. When giving the final predicted credit limit, the risk - bearing capacity of the credit - granting institution is also fully considered, and the final predicted credit limit value is obtained by combining the capital return rate of the lending platform, thereby improving the accuracy of the predicted credit limit value of this application.

[0100] Device Embodiment 1

[0101] An embodiment of the present invention provides an electronic device, including:

[0102] A processor; and,

[0103] A memory arranged to store computer - executable instructions, and when the computer - executable instructions are executed, the processor executes the steps of the above - mentioned method embodiment.

[0104] Device Embodiment 2

[0105] An embodiment of the present invention provides a storage medium for storing computer - executable instructions, and when the computer - executable instructions are executed, the steps of the above - mentioned method embodiment are implemented.

[0106] Finally, it should be noted that: the above - mentioned embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting a user's credit limit, characterized in that Including: S1. Obtain the first credit investigation data and the second credit investigation data of the user based on the user's loan request, obtain a credit granting index according to the first credit investigation data and the second credit investigation data, and determine whether the user is eligible for credit granting based on the credit granting index; S2. If the user is not eligible for credit granting, reject the credit granting request. If the user is eligible for credit granting, obtain the user's income and expenditure flow data and the default probability; S3. Obtain the initial credit granting amount prediction value of the user according to the income and expenditure flow data by using a pre-constructed amount prediction model, and obtain the final credit granting amount prediction value based on the initial credit granting amount prediction value, the default probability, and the capital return rate of the lending platform.

2. The method according to claim 1, characterized in that The specific content of S1 includes: S11. Score according to the first credit investigation data by using a preset first preset condition to obtain a first credit score. If the first credit score is less than the first preset credit value, reject the credit granting request; S12. If the first credit score is greater than the first preset credit value and less than the second preset credit value, obtain the second credit investigation data, score according to the second credit investigation data by using a preset second preset condition to obtain a second credit score, obtain the sum of the first credit score and the second credit score, and determine whether the user is eligible for credit granting; S13. If the first credit score is greater than the second preset credit value, obtain the credit granting qualification.

3. The method according to claim 2, wherein The specific content of obtaining the sum of the first credit score and the second credit score and determining whether the user is eligible for credit granting includes: Compare the sum of the first credit score and the second credit score with the first preset credit value. If the sum of the first credit score and the second credit score is greater than the first preset credit value, obtain the credit granting qualification; otherwise, reject the credit granting request.

4. The method according to claim 1, characterized in that The specific content of S11 includes: Obtain the user's credit report data from the People's Bank of China and related persons. The user's credit report data from the People's Bank of China includes basic data and classified data. The basic data includes age, education level, educational background, marital status, and work information; According to the classified data, divide the classified data corresponding to each basic data into a first-level classification, a second-level classification corresponding to the first-level classification, and a feature field corresponding to the second-level classification; Among them, the first-level classification includes account type, business type, and account status. The second-level classification includes non-revolving loan accounts, revolving loan accounts, and credit card accounts under the account type, housing commercial loans, housing provident fund loans, and consumer loans under the business type, overdue accounts, settled, and unsettled under the account status. The feature fields include: original variables and derived variables. Among them, the original variables include: variables mentioned in the user's credit report data from the People's Bank of China, including monthly consumption records of each credit card, newly opened accounts, car loan information, and mortgage information. The derived variables are new variables calculated based on the original variables, including the total credit card consumption amount and the total number of accounts; After processing the outlier and missing value information of the divided credit report data from the People's Bank of China, score the user's credit report data from the People's Bank of China to obtain the first credit score.

5. The method according to claim 1, wherein The second credit investigation data is the first credit investigation data of the related person.

6. The method according to claim 1, characterized in that, The pre-constructed amount prediction model is constructed through the following steps: Obtain modeling variables, specifically including: performing feature derivation based on the income and expenditure transaction data and loan request data of historical credit-granted users to obtain feature-derived variables, where the feature-derived variables include: the average monthly deposit balance of historical credit-granted users, the tax expenditure amount of historical credit-granted users, the operating income amount of historical credit-granted users, and the transaction expenditure amount of historical credit-granted users; determining the loan data information of the historical credit-granted users according to the income and expenditure transaction data of the historical credit-granted users; performing a first screening on the feature-derived variables according to the feature missing rate of the feature-derived variables and the correlation coefficient between the feature-derived variables and the loan data information; performing a second screening on the feature-derived variables obtained after the first screening according to business implications to obtain the modeling variables; Obtain a credit limit prediction model, specifically including: screening a preset number of historical credit-granted users according to preset screening conditions to obtain a training set of historical credit-granted users and a test set of historical credit-granted users, and training a linear regression model established by the Lasso regression method on the modeling variables of the training set and the test set to obtain a credit limit model; The specific process of screening a preset number of historical credit-granted users according to preset screening conditions to obtain a training set of historical credit-granted users and a test set of historical credit-granted users includes: screening historical credit-granted users according to historical credit-granted user screening conditions to obtain a sample of historical credit-granted users, where the historical credit-granted user screening conditions include: the loan date of historical credit-granted users, the scale of historical credit-granted users, the loan amount of historical credit-granted users, the loan data information of historical credit-granted users, outliers of loan data information, and bad credit; determining the cross-time validation set of the historical credit-granted users and the union of the training set and the test set according to a preset time point; determining the training set of historical credit-granted users and the test set of historical credit-granted users in the union according to a preset ratio.

7. The method according to claim 1, characterized in that, The predicted credit limit value of the user includes: credit limit, repayment method, repayment term, and loan interest rate.

8. The method according to claim 1, wherein The specific process of obtaining the final predicted credit limit value based on the initial predicted credit limit value and the default probability in combination with the capital return rate of the lending platform includes: Calculating the capital return rate of the loan according to the loan policy, interest rate setting, and operating cost of the lending platform; Taking the initial predicted credit limit value, default probability, and capital return rate as input variables, inputting them into a comprehensive evaluation model, and setting different weights according to the risk and return preferences of the lending platform to obtain the final predicted credit limit value.

9. An electronic device, including: A processor; And, A memory arranged to store computer-executable instructions, and when the computer-executable instructions are executed, the processor executes the steps of the method for predicting the user's credit limit according to any one of claims 1-8.

10. A storage medium for storing computer-executable instructions, and when the computer-executable instructions are executed, the steps of the method for predicting the user's credit limit according to any one of claims 1-8 are implemented.

Images