Financial risk control prediction model construction method

Abstract

The invention discloses a financial risk control prediction model construction method, and relates to the technical field of financial risk control prediction, a data collection module is utilized to collect and store features related to an evaluation period of a current customer, and a customer feature evaluation module is utilized to construct a prediction model. Sequentially calculating and outputting a risk score F, a risk early warning index FY and a dynamic risk adjustment coefficient DT, and evaluating and updating an interest rate adjustment coefficient V of the current customer in the next evaluation period by using a customer feature evaluation module to obtain a risk evaluation result; according to the invention, multi-dimensional risk factors are comprehensively considered, root operation is introduced to enhance a non-linear relationship, an interest rate adjustment coefficient V is flexibly adjusted, historical and current risk information is combined, overdue severity is quantified, comprehensive influence of a credit score P and repayment capability HN is considered, a latest credit score P is introduced to carry out dynamic adjustment, and a cyclic influence mechanism is formed. The innovations aim at improving the accuracy and flexibility of risk prediction and providing more effective risk control support for financial institutions.

Description

Technical Field

[0001] The present invention relates to the technical field of financial risk control prediction, and in particular to a method for constructing a financial risk control prediction model. Background Art

[0002] With the rapid development of financial technology, financial institutions have an increasing demand for risk management. Traditional risk management methods mainly rely on manual review and simple statistical models, which are difficult to cope with large-scale, high-frequency transactions and complex market environments. Therefore, financial institutions have begun to adopt advanced financial risk control prediction models to improve the efficiency and accuracy of risk management. The widespread application of big data and artificial intelligence technologies has provided strong support for the construction of financial risk control prediction models. Financial institutions can collect and integrate data from different channels to form a huge data set. Through big data analysis and artificial intelligence technology, financial institutions can process and analyze this data in real time to uncover potential risk points and fraudulent behaviors.

[0003] However, some of the existing technologies currently rely mainly on static data for analysis, but ignore the dynamic changes in customer credit status, resulting in inaccurate risk assessment results. Some existing risk control models are usually based on fixed rules and linear relationships for prediction, which makes it difficult to adapt to the complex and changing market environment and customer credit status. In addition, the existing technologies lack an effective risk warning mechanism and are unable to timely predict and identify potential risk events. In addition, a large amount of useful customer information is not fully utilized, resulting in limited predictive capabilities of risk control models. Summary of the invention

[0004] The purpose of the present invention is to provide a method for constructing a financial risk control prediction model, which solves the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solution, and the specific implementation steps are as follows: Step 1: Use the data collection module to collect and store the characteristics related to the current customer's evaluation cycle; Step 2: Use the prediction model constructed by the customer feature assessment module to calculate the output risk score F, risk warning index FY, and dynamic risk adjustment coefficient DT in sequence; Step 3: Based on the dynamic risk adjustment coefficient DT and using the customer feature evaluation module, evaluate and update the interest rate adjustment coefficient V of the current customer in the next evaluation period; Step 4: using the monitoring module and the customer feature evaluation module to regularly monitor and update the evaluation prediction output by the customer feature evaluation module; The customer characteristics assessment module includes a unit for measuring the customer's credit risk level, a unit for predicting the degree of risk faced by the customer in the future, and a unit for adjusting the risk based on the customer's latest credit; The devices used by the data collection module include storage devices; The equipment used by the customer feature evaluation module includes a server; The equipment used by the monitoring module includes monitoring tools.

[0006] Optionally, the calculation formula for the unit for measuring the customer credit risk level is as follows: ; in: F is the risk score; DE is the loan amount, DE represents the total loan amount applied for by the customer; YC is the number of overdue payments. YC indicates the number of overdue payments made by the customer during this evaluation period. WD is the number of outstanding loans. WD indicates the number of outstanding loans currently held by the customer. P is the credit score; V is the interest rate adjustment factor.

[0007] Optionally, the calculation formula of the credit score P is as follows: ; ; HC is the total number of repayments, which means the total number of repayments that the customer needs to make during this evaluation period; YT is the number of days overdue, which means the total number of days the customer is overdue on all loans; Q is the loan term, Q represents the total term of the customer’s current loan amount DE; T avg is the average repayment period in days; n is the repayment period; T 1 is the repayment days of the first cycle, T 2 is the repayment days for the second cycle, T 3 is the repayment days for the third cycle, T n is the number of repayment days in the nth cycle; Z is the total client assets; A1, A2 and A3 are weight coefficients; A1 focuses on the punctuality of repayment record; A2 focuses on the length of overdue days; A3 focuses on the matching degree between loan amount DE and customer’s financial capacity; A1, A2 and A3 are set according to the risk preferences and business strategies of the financial institutions and satisfy the condition of A1+A2+A3=1. If the risk preferences and business strategies of the financial institutions are ranked as A2>A1>A3, then A1 is 0.3, A2 is 0.5 and A3 is 0.2.

[0008] Optionally, the calculation formula of the unit for predicting the degree of risk faced by the customer in the future is as follows: ; ; in: FY is the risk warning index; HN is repayment capacity; IN is the customer’s monthly income; I0 is the monthly expenditure of the customer; YC / DE represents the ratio of the number of overdue payments YC to the loan amount DE, which is used to measure the severity of the overdue payment; It represents the impact of the interest rate adjustment coefficient V and the number of outstanding loans WD on the risk warning index FY. At the same time, the credit score P and repayment capacity HN are introduced to conduct a comprehensive assessment of the impact.

[0009] Optionally, the calculation formula for adjusting the risk unit based on the customer's latest credit is as follows: ; in: DT is the dynamic risk adjustment factor; P lasttime Credit score for the last assessment cycle; YC max is the maximum number of overdue payments, YC max It indicates the maximum number of overdue payments allowed by the financial institution during the assessment period. It indicates the influence of the change of credit score P on the dynamic risk adjustment coefficient DT; Indicates the risk score F, the number of overdue payments YC relative to the maximum number of overdue payments YC max The proportion of and the impact of the loan amount DE on the dynamic risk adjustment coefficient DT.

[0010] Optionally, based on the dynamic risk adjustment coefficient DT, the adjustment calculation formula for the interest rate adjustment coefficient V of the next evaluation period is as follows: ; V new is the adjusted interest rate adjustment factor; a is the adjustment coefficient, which represents the influence of the dynamic risk adjustment coefficient DT on the interest rate adjustment coefficient V, and this coefficient needs to be set according to the risk preferences and business strategies of the financial institution.

[0011] Optionally, the adjustment coefficient a is set as follows: If a financial institution needs to control risks more strictly, the adjustment coefficient a shall be within the range of {0.5≤a≤1.0}; If a financial institution needs to flexibly adjust interest rates to attract customers, the adjustment coefficient a should be in the range of {0.01≤a≤0.5}.

[0012] Optionally, the monitoring module regularly collects the same loan amount DE of m customers and the risk score F in the same evaluation cycle, and uses the customer feature evaluation module to calculate and evaluate the accuracy, recall rate and F1 score indicators of the constructed prediction model in sequence; Among them, the risk score F is divided into the risk score F calculated and output by the constructed prediction model, and the risk score F that has occurred and is calculated and output based on the actual situation.

[0013] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention introduces the change of credit score P, repayment capacity HN and dynamic risk adjustment coefficient DT factors, so that the constructed prediction model can more comprehensively evaluate the credit risk level of the customer, thereby improving the accuracy of risk assessment. Specifically, the result value risk score F of the unit for measuring the credit risk level of the customer comprehensively considers the loan amount DE, the number of overdue payments YC, the number of outstanding loans WD and the credit score P factors, and combines static data with dynamic factors through square root operation and multiplication operation to obtain a more accurate risk score F. The result value risk warning index FY of the unit for predicting the degree of risk faced by the customer in the future further considers the ratio of the number of overdue payments YC to the loan amount DE and the influence of the interest rate adjustment coefficient V and the number of outstanding loans WD on the risk warning index FY. The accuracy of risk warning is improved through weighted and square root operation. Finally, the result value dynamic risk adjustment coefficient DT of the risk adjustment unit according to the customer's latest credit adjusts the risk level of the loan according to the customer's latest credit status. By introducing the change of credit score P and the number of overdue payments YC relative to the maximum number of overdue payments YC max The proportional factor realizes the dynamic adjustment and prediction of risks.

[0014] 2. The prediction model constructed by the present invention adopts nonlinear relationships for prediction, which can adapt to the complex and changing market environment and customer credit status more flexibly. Through the cyclical influence mechanism, the constructed prediction model can adjust the risk assessment results in real time according to the changes in the customer's credit status, thereby improving the flexibility and adaptability of the model.

[0015] 3. The present invention calculates the risk warning index FY so that the constructed prediction model can predict and identify potential risk events in advance and provide timely risk warning information for financial institutions.

[0016] 4. The prediction model constructed by the present invention makes full use of the useful information of the customer's loan amount DE, number of overdue payments YC, number of outstanding loans WD, credit score P, repayment ability HN and the latest credit score, thereby improving the data utilization rate and the prediction ability of the risk control model. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flowchart of the method for constructing the financial risk control prediction model; Figure 2 This is a flow chart of the monitoring and evaluation method in the financial risk control prediction model construction method; Figure 3 It is a structural schematic diagram of the customer characteristic evaluation module of the present invention; Figure 4 Schematic diagram of risk adjustment within the circular feedback cycle of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Regarding the construction method of this financial risk control prediction model, it is different from the existing financial risk control prediction model construction method. The existing financial risk control prediction model construction method ignores the dynamic changes of customer credit status and is difficult to adapt to the complex and changeable market environment and customer credit status. In addition, the existing technology lacks an effective risk warning mechanism, which leads to limited prediction ability of the risk control model. This algorithm unit comprehensively considers multi-dimensional risk factors, introduces square root operations to enhance nonlinear relationships, flexibly adjusts the interest rate adjustment coefficient V, combines historical and current risk information, quantifies the severity of overdue, considers the comprehensive impact of credit score P and repayment ability HN, introduces the latest credit score P for dynamic adjustment, and forms a circular influence mechanism. These innovations are aimed at improving the accuracy and flexibility of risk prediction and providing financial institutions with more effective risk control support.

[0020] For example, see Figures 1 to 4 This implementation provides a method for building a financial risk control prediction model. The specific implementation steps are as follows: Step 1: Use the data collection module to collect and store the characteristics related to the current customer's evaluation cycle; Step 2: Use the prediction model constructed by the customer feature assessment module to calculate the output risk score F, risk warning index FY, and dynamic risk adjustment coefficient DT in sequence; Step 3: Based on the dynamic risk adjustment coefficient DT and using the customer feature evaluation module, evaluate and update the interest rate adjustment coefficient V for the next evaluation period of the current customer; Step 4: Use the monitoring module and the customer feature evaluation module to regularly monitor and update the evaluation forecast output by the customer feature evaluation module; Among them, the customer characteristics assessment module includes a unit for measuring the customer's credit risk level, a unit for predicting the degree of risk faced by the customer in the future, and a unit for adjusting the risk based on the customer's latest credit; The devices used by the data collection module include storage devices; The equipment used in the customer characteristics assessment module includes a server; The equipment used by the monitoring module includes monitoring tools; The monitoring module regularly collects the same loan amount DE of m customers and the risk score F in the same assessment cycle, and uses the customer feature assessment module to calculate and evaluate the accuracy, recall rate, and F1 score indicators of the constructed prediction model in turn; Among them, the risk score F is divided into the risk score F calculated and output by the constructed prediction model, and the risk score F that has occurred and is calculated and output based on the actual situation.

[0021] In this embodiment, the system, through the mutual cooperation of three algorithm units, as well as different methods and combinations, jointly affects the calculation results of the final risk score F, the risk warning index FY and the dynamic risk adjustment coefficient DT, and combines the three calculation results of F, FY and DT to provide financial institutions with more accurate and comprehensive customer credit risk assessment information, thereby helping financial institutions make more wise loan decisions. Specifically, F is the risk score, and the risk score F is an important basis for financial institutions to conduct risk assessment and loan approval, and this algorithm unit can generate a relatively accurate risk score F to help financial institutions judge the credit risk of customers. FY is the risk warning index, and the index can more comprehensively reflect the credit status and repayment ability HN of customers, providing financial institutions with Provide earlier risk warning signals so that they can take corresponding risk management measures in advance. DT is the dynamic risk adjustment coefficient, which can reflect changes in customer credit status in real time and dynamically adjust the risk level of the loan according to these changes, making the risk management of financial institutions more flexible and accurate. The calculation results of DT can also affect the calculation of F and FY, so that the three algorithms of this system together constitute a complete risk assessment system and provide financial institutions with comprehensive, accurate and flexible risk management tools. Through the comprehensive use of these three algorithm units, financial institutions can more accurately assess the credit risk level of customers, predict potential risk events, and dynamically adjust the risk level of loans according to the latest credit status of customers, thereby optimizing risk management strategies and improving risk management effects.

[0022] See also Figures 1 to 4 , the calculation formula for measuring the customer credit risk level unit is as follows: ; in: F is the risk score; DE is the loan amount, DE represents the total loan amount applied for by the customer; YC is the number of overdue payments. YC indicates the number of overdue payments made by the customer during this evaluation period. WD is the number of outstanding loans. WD indicates the number of outstanding loans currently held by the customer. P is the credit score; V is the interest rate adjustment factor.

[0023] In this embodiment: First, the calculation part "[DE×(YC+WD)" in this algorithm unit is intended to comprehensively consider the impact of the loan amount DE, the number of overdue payments YC and the number of outstanding loans WD on the risk score. The larger the loan amount DE, the more overdue payments TC and the more outstanding loans WD, the higher the credit risk of the customer. As an important factor in the calculation of the risk score F, this calculation part combines the three key indicators of the loan amount DE, the number of overdue payments YC and the number of outstanding loans WD in a weighted manner to jointly affect the final risk score result; The calculation part of "SQRT(P)" takes the square root of the credit score P, which can smooth the changes of the credit score P and make the differences between different credit scores P more reasonable. At the same time, the square root operation can also maintain the positive impact of the credit score P on the risk score F. As a multiplier factor in the calculation of the risk score F, the credit score P after the square root can more accurately reflect the impact of the customer's credit status on the risk score. In addition, the interest rate adjustment coefficient V is used as a subtrahend to reflect the direct impact of the interest rate adjustment on the risk score F. This algorithm unit adjusts the value of the interest rate adjustment coefficient V, allowing financial institutions to flexibly adjust risk scores according to changes in the market environment and customer credit status, thereby more accurately reflecting the customer's credit risk level. The introduction of the interest rate adjustment coefficient V enables the risk score F to be flexibly adjusted according to changes in market interest rates and customer credit status, thereby improving the timeliness and accuracy of risk assessment. The expression for measuring the customer credit risk level is concise and easy to understand and apply, which is convenient for financial institutions to calculate the risk score F in actual business. The square root operation "SQRT(P)" is used in it, which enhances the model's nonlinear processing ability for the credit score P. In actual applications, the relationship between the credit score P and the risk level is often nonlinear, so this processing method can improve the accuracy of the model. Based on the above, the unit for measuring customer credit risk level can more comprehensively assess the customer's credit risk level and avoid assessment bias caused by a single indicator.

[0024] See also Figures 1 to 4 , the calculation formula for predicting the risk level of customers in the future is as follows: ; ; in: FY is the risk warning index; HN is repayment capacity; IN is the customer’s monthly income; I0 is the monthly expenditure of the customer; YC / DE represents the ratio of the number of overdue payments YC to the loan amount DE, which is used to measure the severity of the overdue payment; It represents the impact of the interest rate adjustment coefficient V and the number of outstanding loans WD on the risk warning index FY. At the same time, the credit score P and repayment capacity HN are introduced to conduct a comprehensive assessment of the impact.

[0025] In this embodiment, the "YC / DE" calculation part first calculates the ratio of the number of overdue payments YC to the loan amount DE, which is used to measure the severity of the overdue. The higher this ratio is, the higher the credit risk of the customer is. As a multiplier factor in the calculation of the risk warning index FY, "YC / DE" can more accurately reflect the impact of the number of overdue payments YC on the risk warning index FY. The calculation part aims to comprehensively consider the impact of credit score P, repayment capacity HN, interest rate adjustment coefficient V and outstanding loan number WD on risk warning index FY. Through square root operation and smoothing, this calculation part can more accurately reflect the overall credit status and repayment ability of the customer. As a subtractive factor in the calculation of risk warning index FY, this calculation part combines the four key indicators of credit score P, repayment capacity HN, interest rate adjustment coefficient V and outstanding loan number WD through weighted and smoothing, thereby jointly affecting the final risk warning index FY result; In this embodiment, this algorithm unit can give early warning of risk events that customers may face in the future by comprehensively considering multiple factors, providing financial institutions with sufficient time to respond to risks, and the interest rate adjustment coefficient V and the outstanding loan amount WD factors in the unit for predicting the degree of risk faced by customers in the future enable risk assessment to be dynamically adjusted as the credit status of customers changes, thereby improving the flexibility and accuracy of risk assessment; In addition, by introducing the key factor of repayment capacity HN, the unit for predicting the degree of risk faced by customers in the future can more accurately assess the ability of customers to repay, thereby enhancing the reliability of risk warning; It is worth noting that in the unit for predicting the degree of risk that customers will face in the future, the severity of overdue payments is quantified by calculating the ratio of the number of overdue payments YC to the loan amount DE. This quantification method enables the model to more accurately assess the impact of overdue payments on customer credit risk. In addition, the comprehensive impact of credit score P and repayment ability HN on the risk warning index FY is also considered. This processing method enables the model to more comprehensively assess the impact of customers' repayment ability and credit status on risk warning.

[0026] See also Figures 1 to 4 , the calculation formula for adjusting the risk unit based on the customer's latest credit is as follows: ; in: DT is the dynamic risk adjustment factor; P lasttime Credit score for the last assessment cycle; YC max is the maximum number of overdue payments, YC max It indicates the maximum number of overdue payments allowed by the financial institution during the assessment period. It indicates the influence of the change of credit score P on the dynamic risk adjustment coefficient DT; Indicates the risk score F, the number of overdue payments YC relative to the maximum number of overdue payments YC max The proportion of and the impact of the loan amount DE on the dynamic risk adjustment coefficient DT.

[0027] In this embodiment, the algorithm unit first The calculation part aims to measure the change of credit score P by comparing the latest credit score P with the credit score P of the previous assessment cycle. lasttime , which can reflect the changes in the customer's credit status, and as a multiplier factor in the calculation of the dynamic risk adjustment coefficient DT, this calculation part can more accurately reflect the impact of changes in the customer's credit status on the dynamic risk adjustment coefficient DT; The calculation part aims to comprehensively consider the loan amount DE, risk score F and the number of overdue payments YC relative to the maximum number of overdue payments YC max The impact of the proportion of on the dynamic risk adjustment coefficient DT. By adjusting the value of this calculation part, it can more flexibly reflect the changes in customer credit risk. As an addend factor in the calculation of the dynamic risk adjustment coefficient DT, this calculation part combines the three key indicators of loan amount DE, risk score F and number of overdue payments YC through weighted and proportional adjustment, and jointly affects the final dynamic risk adjustment coefficient DT result; In this algorithm unit, the latest credit score P and the credit score P of the previous evaluation cycle are introduced. lasttime Key factors enable the risk unit adjusted according to the customer's latest credit to reflect changes in the customer's credit status in real time, provide financial institutions with the latest risk assessment results, and by comprehensively considering multiple factors, the risk unit adjusted according to the customer's latest credit can more accurately assess the customer's risk level, and dynamically adjust the risk level of the loan based on the assessment results, thereby optimizing the effect of risk management; It is worth noting that the risk unit is calculated based on the customer's latest credit adjustment. , and then consider the impact of changes in credit score P on the dynamic risk adjustment coefficient DT. This processing method enables the model to more comprehensively evaluate the impact of historical changes in customer credit status and current status on risk. The result value DT of the customer's latest credit adjustment risk unit can be used to adjust the interest rate adjustment coefficient V in the unit that measures the customer's credit risk level, thereby forming a circular influence mechanism. This mechanism enables the model to more flexibly adapt to changes in customer credit status and improve the accuracy and flexibility of risk prediction; To sum up, the innovations in the methods of constructing financial risk control prediction models, namely, the unit for measuring customer credit risk level, the unit for predicting the degree of risk faced by customers in the future, and the unit for adjusting risk based on customers' latest credit, are mainly reflected in the comprehensive consideration of multi-dimensional risk factors, the introduction of square root operations to enhance nonlinear relationships, the flexible adjustment of interest rate adjustment coefficients, the combination of historical and current risk information, the quantification of the severity of overdue payments, the consideration of the combined impact of credit ratings and repayment ability, the introduction of the latest credit ratings for dynamic adjustment, and the formation of a cyclical influence mechanism. These innovations are aimed at improving the accuracy and flexibility of risk prediction and providing financial institutions with more effective risk control support.

[0028] See also Figures 1 to 4 , based on the dynamic risk adjustment coefficient DT, the adjustment calculation formula for the interest rate adjustment coefficient V of the next assessment period is as follows: ; V new is the adjusted interest rate adjustment factor; a is the adjustment coefficient, which indicates the influence of the dynamic risk adjustment coefficient DT on the interest rate adjustment coefficient V, and this coefficient needs to be set according to the risk preference and business strategy of the financial institution; The setting of adjustment coefficient a is as follows: If a financial institution needs to control risks more strictly, the adjustment coefficient a shall be within the range of {0.5≤a≤1.0}; If a financial institution needs to flexibly adjust interest rates to attract customers, the adjustment coefficient a should be in the range of {0.01≤a≤0.5}.

[0029] In this embodiment, the algorithm unit uses a cyclic influence mechanism to enable the constructed prediction model to adjust the risk assessment results and the risk level of the loan in real time according to changes in the customer's credit status, thereby enhancing the flexibility of the model; Specifically, the cyclic impact mechanism enables the model to comprehensively consider data and factors at multiple time points, so as to more accurately assess the customer's credit risk level. Through the cyclic impact mechanism, financial institutions can dynamically adjust risk management strategies based on the latest risk assessment results to better deal with potential risks. When financial institutions want to control risks more strictly, they can choose a large adjustment coefficient a. In this way, a small change in the dynamic risk adjustment coefficient DT will lead to a significant change in the interest rate adjustment coefficient V. This setting helps financial institutions to respond quickly when risk signals first appear and reduce potential losses by adjusting interest rates. On the contrary, when financial institutions want to adjust interest rates more flexibly to attract customers, they can choose a small adjustment coefficient a. In this way, even if the dynamic risk adjustment coefficient DT changes significantly, the change in the interest rate adjustment coefficient V is relatively limited. This setting enables financial institutions to respond more flexibly to changes in market competition and customer demand while maintaining a certain level of risk control. By setting a reasonable adjustment coefficient a, financial institutions can set prices more accurately based on the credit status and risk level of customers, which helps financial institutions improve customer satisfaction and loyalty while maintaining profitability. The setting of adjustment coefficient a can also help financial institutions find a balance between risk and return. A large adjustment coefficient a means higher risk control requirements, but it will also lead to the loss of some high-quality customers, while a small adjustment coefficient a helps attract more customers, but also increases potential risks. Therefore, financial institutions need to choose a suitable adjustment coefficient a based on their own situation and market environment, as well as the value range of adjustment coefficient a. By combining the dynamic risk adjustment coefficient DT with the adjustment coefficient a, financial institutions can realize the automation and intelligence of risk management. In this way, financial institutions can more efficiently monitor and evaluate the credit risk status of customers and take corresponding measures to reduce risks in a timely manner. Automated risk management can also help financial institutions reduce operating costs. By reducing manual intervention and error rates, financial institutions can more efficiently process large amounts of customer information and reduce losses caused by improper risk management. Reasonable setting of adjustment coefficient a can help financial institutions promote product innovation. By flexibly adjusting interest rate pricing strategies, financial institutions can develop financial products that better meet market demand and customer preferences, thereby enhancing market competitiveness. In addition, reasonable setting of adjustment coefficient a can also provide strong support for the business development of financial institutions. By optimizing interest rate pricing strategies, financial institutions can attract more high-quality customers, expand market share, and promote sustained business growth. To sum up, the value range setting of the adjustment coefficient α and the adjustment calculation formula combined with the interest rate adjustment coefficient V have many beneficial effects for financial institutions. These effects not only help financial institutions enhance the flexibility of risk control, optimize interest rate pricing strategies, and improve risk management efficiency, but also promote business innovation and development. Therefore, in actual applications, financial institutions need to select appropriate adjustment coefficient a according to their own situation and market environment, and continuously optimize and adjust interest rate pricing strategies to cope with market changes.

[0030] For example 2, please refer to Figures 1 to 4 , the calculation formula of credit score P is as follows: ; ; HC is the total number of repayments, which means the total number of repayments that the customer needs to make during this evaluation period; YT is the number of days overdue, which means the total number of days the customer is overdue on all loans; Q is the loan term, Q represents the total term of the customer’s current loan amount DE; T avg is the average repayment period in days; n is the repayment period; T 1 is the repayment days of the first cycle, T 2 is the repayment days for the second cycle, T 3 is the repayment days for the third cycle, T n is the number of repayment days in the nth cycle; Z is the total client assets; A1, A2 and A3 are weight coefficients; A1 focuses on the punctuality of repayment record; A2 focuses on the length of overdue days; A3 focuses on the matching degree between loan amount DE and customer’s financial capacity; A1, A2 and A3 are set according to the risk preferences and business strategies of the financial institutions and satisfy the condition of A1+A2+A3=1. If the risk preferences and business strategies of the financial institutions are ranked as A2>A1>A3, then A1 is 0.3, A2 is 0.5 and A3 is 0.2.

[0031] In this embodiment, A1 focuses on the punctuality of repayment records. This proportion is calculated by the logarithmic function, which aims to more accurately reflect the negative impact of the number of overdue payments YC on the credit score P. Therefore, the larger the A1, the more the financial institution values ​​the customer's repayment punctuality. If the customer has a good repayment record and few overdue payments, the score of this part will be higher, thereby improving the overall credit score. A2 focuses on the length of overdue days. This ratio reflects the ratio of the length of the customer's overdue time to the loan term Q. The setting of A2 reflects the financial institution's attention to the length of the overdue days YT. If the customer is overdue but the number of days is short, then the score of this part may not be too low. On the contrary, if the number of overdue days is long, then the score will be significantly reduced. A3 focuses on the matching degree between the loan amount and the customer's financial ability. A3 is related to the inverse of "(DE / Z)". This ratio reflects whether the customer's loan amount matches his or her financial ability. The setting of A3 shows the importance that financial institutions attach to the relationship between the loan amount and the customer's actual financial ability. If the customer's loan amount is too high and exceeds his or her financial ability, the score of this part will be low. On the contrary, if the loan amount is moderate and matches the customer's financial ability, the score will be high. It should be noted that although A1, A2 and A3 each focus on different aspects, their roles in the comprehensive credit score are interrelated. Financial institutions need to consider various factors when setting these weight coefficients to ensure that the credit score can fully and accurately reflect the customer's credit status. At the same time, these weight coefficients also need to be adjusted in a timely manner according to changes in the market environment and business strategies.

[0032] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for constructing a financial risk control prediction model, characterized in that: The specific implementation steps are as follows: Step 1: Use the data collection module to collect and store the characteristics related to the current customer's evaluation cycle; Step 2: Use the prediction model constructed by the customer feature assessment module to calculate the output risk score F, risk warning index FY, and dynamic risk adjustment coefficient DT in sequence; Step 3: Based on the dynamic risk adjustment coefficient DT and using the customer feature evaluation module, evaluate and update the interest rate adjustment coefficient V of the current customer in the next evaluation period; Step 4: using the monitoring module and the customer feature evaluation module to regularly monitor and update the evaluation prediction output by the customer feature evaluation module; The customer feature assessment module includes a unit for measuring the customer's credit risk level, a unit for predicting the degree of risk that the customer will face in the future, and a unit for adjusting the risk based on the customer's latest credit.

2. A method for constructing a financial risk control prediction model according to claim 1, characterized in that: The devices used by the data collection module include storage devices; The equipment used by the customer feature evaluation module includes a server; The equipment used by the monitoring module includes monitoring tools.

3. A method for constructing a financial risk control prediction model according to claim 2, characterized in that: The calculation formula for measuring the customer credit risk level unit is as follows: ； in: F is the risk score; DE is the loan amount, DE represents the total loan amount applied for by the customer; YC is the number of overdue payments. YC indicates the number of overdue payments made by the customer during this evaluation period. WD is the number of outstanding loans. WD indicates the number of outstanding loans currently held by the customer. P is the credit score; V is the interest rate adjustment factor.

4. A method for constructing a financial risk control prediction model according to claim 3, characterized in that: The calculation formula of the credit score P is as follows: ； ； HC is the total number of repayments, which means the total number of repayments that the customer needs to make during this evaluation period; YT is the number of days overdue, which means the total number of days the customer is overdue on all loans; Q is the loan term, Q represents the total term of the customer’s current loan amount DE; T avg is the average repayment period in days; n is the repayment period; T1 is the repayment days for the first cycle, T2 is the repayment days for the second cycle, and T3 is the repayment days for the third cycle. n is the number of repayment days in the nth cycle; Z is the total client assets; A1, A2 and A3 are weight coefficients; A1 focuses on the punctuality of repayment record; A2 focuses on the length of overdue days; A3 focuses on the matching degree between loan amount DE and customer’s financial capacity; The calculation formula for predicting the risk level of customers in the future is as follows: ； ； in: FY is the risk warning index; HN is repayment capacity; IN is the customer’s monthly income; I0 is the monthly expenditure of the customer; YC / DE represents the ratio of the number of overdue payments YC to the loan amount DE, which is used to measure the severity of the overdue payment; It represents the impact of the interest rate adjustment coefficient V and the number of outstanding loans WD on the risk warning index FY. At the same time, the credit score P and repayment capacity HN are introduced to conduct a comprehensive assessment of the impact.

5. A method for constructing a financial risk control prediction model according to claim 4, characterized in that: In the calculation formula of the credit score P, A1, A2 and A3 are set according to the risk preference and business strategy of the financial institution, and satisfy the condition of A1+A2+A3=1. If the risk preference and business strategy of the financial institution are ranked as A2>A1>A3, then A1 is 0.3, A2 is 0.5, and A3 is 0.

2.

6. A method for constructing a financial risk control prediction model according to claim 5, characterized in that: The calculation formula for adjusting the risk unit based on the customer's latest credit is as follows: ； in: DT is the dynamic risk adjustment factor; P lasttime Credit score for the last assessment cycle; YC max is the maximum number of overdue payments, YC max It indicates the maximum number of overdue payments allowed by the financial institution during the assessment period. It indicates the influence of the change of credit score P on the dynamic risk adjustment coefficient DT; Indicates the risk score F, the number of overdue payments YC relative to the maximum number of overdue payments YC max The proportion of and the impact of the loan amount DE on the dynamic risk adjustment coefficient DT.

7. A method for constructing a financial risk control prediction model according to claim 6, characterized in that: Based on the dynamic risk adjustment coefficient DT, the adjustment calculation formula for the interest rate adjustment coefficient V of the next assessment period is as follows: ； V new is the adjusted interest rate adjustment factor; a is the adjustment coefficient, which represents the influence of the dynamic risk adjustment coefficient DT on the interest rate adjustment coefficient V, and this coefficient needs to be set according to the risk preferences and business strategies of the financial institution.

8. A method for constructing a financial risk control prediction model according to claim 7, characterized in that: The setting of the adjustment coefficient a is specifically as follows: If a financial institution needs to control risks more strictly, the adjustment coefficient a shall be within the range of {0.5≤a≤1.0}; If a financial institution needs to flexibly adjust interest rates to attract customers, the adjustment coefficient a should be in the range of {0.01≤a≤0.5}.

9. A method for constructing a financial risk control prediction model according to claim 8, characterized in that: The monitoring module regularly collects the same loan amount DE of m customers and the risk score F in the same evaluation cycle, and uses the customer feature evaluation module to calculate and evaluate the accuracy, recall rate and F1 score indicators of the constructed prediction model in turn; Among them, the risk score F is divided into the risk score F calculated and output by the constructed prediction model, and the risk score F that has occurred and is calculated and output based on the actual situation.

Images