Credit investigation variable combination mining method and system
By combining and deriving and value calculations of the basic credit reporting variables, a high-value credit reporting variable combination is generated, which solves the problem that the potential value of credit reporting data in the existing technology is underutilized, and achieves more accurate distinction between good and bad customers and support for financial decision-making.
Patent Information
- Application Number
- CN202411606287.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-12
- Publication Date
- 2025-02-28
AI Technical Summary
When processing PBOC credit reporting messages, the existing technology lacks in-depth mining and combination analysis, and cannot fully discover the potential value of credit reporting data, resulting in limited ability to distinguish good and bad customers.
By obtaining the customer's historical credit reporting data, extracting the basic credit reporting variables, and combining and deriving these variables using machine learning algorithms and mathematical methods to generate a new combination of credit reporting variables. The value of these combinations is then calculated through a pre-built computing framework, the combinations that meet the criteria are screened, and the KS value and post-combination gain are calculated to evaluate their discriminatory ability and correlation.
In-depth mining of credit reporting data has been achieved, a high-value credit reporting variable combination has been generated, the distinction between good and bad customers has been optimized, and the accuracy of decision-making in financial institutions in credit approval and risk management has been improved.
Smart Images
Figure CN119537893A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of derivation and value mining of risk control features, and in particular to a credit variable combination mining method and system. Background Art
[0002] In the financial sector, credit data is an important basis for assessing the credit status of individuals or enterprises. With the rapid development of financial services, the People's Bank of China's credit reporting system provides a large amount of credit information about individuals and enterprises, including loan records, repayment status, credit card usage, etc. These data are of great value to financial institutions in credit approval, risk management, etc. However, how to efficiently mine and use these credit data to more accurately assess customer credit status has always been a challenge faced by financial institutions.
[0003] At present, many companies mainly adopt the following practices when handling the credit information of the People's Bank of China:
[0004] First, simply analyzing the data and storing it in the database for direct use: This method is relatively simple, mainly parsing the data in the credit report and then directly storing it in the database for subsequent use. Although this approach can retain the original data, it lacks in-depth mining and combined analysis of the data, and cannot fully explore the potential value of credit data.
[0005] Second, ask consulting companies to perform disassembly and statistics: Some companies choose to cooperate with professional consulting companies to disassemble and statistically analyze credit data. Although this method can obtain certain data analysis results, the service cost of consulting companies is high, and the analysis results may not fully meet the personalized needs of the company. In addition, this method still remains at the level of data disassembly and statistics, and fails to conduct deeper combination mining. Summary of the invention
[0006] In view of this, it is necessary to provide a credit variable combination mining method and system to solve the above-mentioned defects of the prior art.
[0007] In order to solve the above problems, in a first aspect, an embodiment of the present invention provides a credit variable combination mining method, comprising:
[0008] Obtain the customer's historical credit data and extract basic credit variables;
[0009] Use machine learning algorithms to combine and derive basic credit variables. For each newly generated credit variable combination, calculate the value of the credit variable combination through a pre-built calculation framework, and screen the credit variable combinations that meet the standards;
[0010] Calculate the KS value of the credit basic variables and the credit variable combination they constitute, and calculate the gain after the combination; the KS value is used to evaluate the ability of the credit basic variables to distinguish between good and bad customers;
[0011] Eliminate combinations whose correlations between derived combinations are greater than the preset correlation threshold, and combinations whose correlations with the original credit basic variables are greater than the preset correlation threshold, and select the target credit variable combination based on the post-combination gain and correlation.
[0012] Preferably, the step of obtaining the historical credit data of the customer and extracting basic credit variables includes:
[0013] Extract historical credit data of multiple customers across platforms;
[0014] Determine the criteria for dividing good and bad customers, and divide customers into good customer samples and bad customer samples based on their historical credit data;
[0015] Extract basic information units used to describe personal credit status from the customer's historical credit data as basic credit variables, and integrate the basic credit variables into a wide table.
[0016] Preferably, the method of using a machine learning algorithm to combine and derive basic credit variables includes:
[0017] Use machine learning algorithms or mathematical methods to combine and derive basic credit variables;
[0018] Among them, the machine learning algorithms include K-means clustering, PCA principal component analysis and xgboost extreme gradient boosting; the mathematical methods include addition combination, subtraction combination, multiplication combination, division combination, square sum combination and trigonometric function combination.
[0019] Preferably, the method of combining and deriving basic credit variables using a machine learning algorithm further includes:
[0020] Use the for loop to traverse and combine the basic credit variables to generate a new credit variable combination.
[0021] Preferably, before calculating the value of the credit variable combination by using the pre-built calculation framework, the method further includes:
[0022] Build IV calculation framework and KS calculation framework;
[0023] Among them, the IV value calculated by the IV calculation framework is used to evaluate the impact of the basic credit variables on good and bad customers; the IV value and KS value are used to jointly judge the value of the basic credit variables.
[0024] Preferably, after calculating the KS value of the credit basic variable and the credit variable combination composed of the basic variable and calculating the combined gain, the method further includes:
[0025] According to the size of the combined gain, the credit variable combinations are divided into low-gain combinations, medium-gain combinations and high-gain combinations;
[0026] Accordingly, the target credit variable combination is selected based on the combined gain and correlation, including:
[0027] Select a high-gain combination whose correlation meets the requirements as the target credit variable combination.
[0028] In a second aspect, an embodiment of the present invention provides a credit variable combination mining system, including:
[0029] The acquisition module is used to obtain the customer's historical credit data and extract the basic credit variables;
[0030] The combination derivation module is used to combine and derive basic credit variables using machine learning algorithms. For each newly generated credit variable combination, the value of the credit variable combination is calculated through a pre-built calculation framework, and the credit variable combination whose value meets the standard is screened;
[0031] The gain calculation module is used to calculate the KS value of the credit investigation basic variables and the credit investigation variable combination they constitute, and calculate the gain after the combination; wherein the KS value is used to evaluate the ability of the credit investigation basic variables to distinguish between good and bad customers;
[0032] The combination selection module is used to eliminate the combinations whose correlation between derived combinations is greater than the preset correlation threshold, and the combinations whose correlation with the original credit basic variables is greater than the preset correlation threshold, and select the target credit variable combination based on the post-combination gain and correlation.
[0033] In a third aspect, the present invention further provides an electronic device, including a memory and a processor, wherein:
[0034] The memory is used to store programs;
[0035] The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the credit variable combination mining method as described in the embodiment of the first aspect of the present invention.
[0036] In a fourth aspect, the present invention further provides a computer-readable storage medium for storing computer-readable programs or instructions, which, when executed by a processor, can implement the steps in the credit variable combination mining method as described in the embodiment of the first aspect of the present invention.
[0037] Compared with the prior art, the credit variable combination mining method and system provided by the present invention has the following beneficial effects:
[0038] 1) The present invention combines and derives basic credit variables through machine learning algorithms and mathematical methods, generates new credit variable combinations, and calculates the values of these combinations, thereby deeply mining the potential value of credit data.
[0039] 2) The present invention realizes automated and intelligent credit variable combination mining by building an IV calculation framework, a KS calculation framework, and using a for loop to traverse and combine basic credit variables. The present invention optimizes the credit data's ability to distinguish good from bad customers by calculating the KS value of the credit variable combination and evaluating the gain after the combination. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] Figure 1 A flow chart of the credit variable combination mining method provided by the present invention;
[0041] Figure 2 A structural block diagram of the credit variable combination mining system provided by the present invention;
[0042] Figure 3 This is a structural block diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION
[0043] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.
[0044] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.
[0045] At present, the existing technology mainly adopts two methods to process the credit information of the People's Bank of China: one is to simply parse the data and store it in the database for subsequent use. Although this method retains the original data, it lacks in-depth mining; the other is to rely on consulting companies to disassemble and count. Although it can obtain certain analysis results, it is costly and difficult to meet personalized needs. Both methods do not conduct combined mining, resulting in the potential value of credit data not being fully explored, and the ability to distinguish between good and bad customers is also limited. Therefore, the existing technology has the defects of insufficient value mining and low differentiation.
[0046] In view of this, the present invention provides a credit variable combination mining method, which combines and derives basic credit variables through machine learning algorithms and mathematical methods, generates new credit variable combinations, and calculates the value of these combinations, thereby deeply mining the potential value of credit data. The following will be explained and introduced through multiple embodiments.
[0047] Figure 1 The flowchart of the credit variable combination mining method provided by the present invention is shown in FIG. Figure 1 The credit variable combination mining method provided by the present invention at least comprises the following steps:
[0048] Step S1, obtain the customer's historical credit data and extract basic credit variables.
[0049] Specifically, first, the historical credit data of multiple customers are extracted across platforms. In this embodiment, the historical credit data of 2,500 customers are extracted from four credit platforms respectively, for a total of 10,000 samples.
[0050] Then, the classification standard of good and bad customers is determined. In this embodiment, "mob6_60+" is used as the classification standard of bad samples. Here, mob represents the age of the account. mob6_60+ means that after the loan is issued for 6 periods, the overdue days exceed 60 days, and the data of this customer is regarded as a bad customer sample. In this embodiment, "mob6_60+" is used as the classification standard of bad samples, which helps to screen out those borrowers who have not repaid for a long time and are seriously overdue, so as to more accurately assess credit risk.
[0051] Furthermore, based on the historical credit data of 10,000 customers, each customer is divided into good customer samples and bad customer samples;
[0052] Finally, basic information units used to describe personal credit status are extracted from the customer's historical credit data as basic credit variables, and the basic credit variables are integrated into a wide table. Here, basic credit variables can include basic personal information (such as name, gender, age, occupation, etc.), credit transaction information (such as loan records, credit card usage, etc.), public information (such as court judgments, administrative penalties, etc.), etc.
[0053] In this embodiment, the same basic credit variables in data from different sources are integrated to form a unified wide table, which is helpful for subsequent data analysis and processing.
[0054] Step S2, using machine learning algorithms to combine and derive basic credit variables. For each newly generated credit variable combination, the value of the credit variable combination is calculated through a pre-built calculation framework, and the credit variable combinations whose values meet the standards are screened.
[0055] Specifically, after obtaining the basic credit variables corresponding to each customer, the basic credit variables are combined and derived using a machine learning algorithm or a mathematical method. Among them, the machine learning algorithm can use K-means clustering, PCA principal component analysis, and xgboost extreme gradient boosting; the mathematical method can use addition combination, subtraction combination, multiplication combination, division combination, square sum combination, and trigonometric function combination. In this embodiment, the basic credit variables are combined in different ways. This can be a combination of two or three or more variables.
[0056] Then, a for loop is used to traverse and combine the basic credit variables to generate a new combination of credit variables. Among them, the for loop is a programming structure used to systematically traverse and perform a series of operations or calculations. The for loop can repeatedly execute a block of code according to a predetermined rule (such as increasing or decreasing the index) until a certain termination condition is met. In the credit variable combination mining method of the present invention, the for loop is used to traverse and combine the basic credit variables to generate a new combination of credit variables. For the selected basic credit variables, the for loop will be combined one by one in a preset manner (such as two-by-two combinations, three-in-a-group, etc.), and subsequent calculation and evaluation steps will be performed on each combination. This embodiment uses a for loop to traverse and combine the basic credit variables, which can improve the efficiency of combination derivation, ensure the systematic exploration of all possible combinations, and facilitate subsequent expansion and modification.
[0057] Furthermore, the value of the credit variable combination is calculated through a pre-built calculation framework, and the credit variable combination whose value meets the standard is screened. It is understandable that before calculating the value of the credit variable combination through the pre-built calculation framework, it is necessary to pre-build the IV calculation framework and the KS calculation framework. The IV (Information Value) calculation framework is a calculation framework for evaluating the predictive ability of variables. It calculates the IV value based on the information value IV indicator to quantify the degree of influence of the variable on the target variable (such as good and bad customers). The higher the IV value, the stronger the predictive ability of the variable. The KS calculation framework is a calculation framework for evaluating the distinguishing ability of the model. It is based on the Kolmogorov-Smirnov (KS) test and measures the degree to which the model distinguishes between good and bad customers by calculating the maximum difference between the cumulative distributions of good and bad samples. The larger the KS value, the stronger the distinguishing ability of the model.
[0058] In the embodiment of the present invention, the IV value is used to evaluate the degree of influence of the basic credit variables on good and bad customers; the KS value is used to evaluate the ability of the basic credit variables to distinguish between good and bad customers. The IV value and the KS value are used to jointly judge the value of the basic credit variables. The "value" here refers to the ability to distinguish between good and bad customers.
[0059] After calculating the IV value and KS value of the credit variable combination, the credit variable combination that meets IV>=0.1and KS>=0.1 is stored in the process table for further analysis and selection.
[0060] The present invention realizes automated and intelligent credit variable combination mining by building an IV calculation framework, a KS calculation framework, and using a for loop to traverse and combine basic credit variables.
[0061] Step S3, calculate the KS value of the credit investigation basic variable and the credit investigation variable combination composed of it, and calculate the gain after combination.
[0062] Specifically, the KS value of the credit variable combination is compared with the KS value of the corresponding credit basic variable, and the calculated percentage gain is the combined gain. By calculating the combined gain, we can understand how much the ability of the credit basic variables to distinguish good and bad customers has been improved compared with the original basic variables.
[0063] Furthermore, after calculating the KS value of the credit basic variable and the credit variable combination composed of it, and calculating the gain after the combination, the credit variable combination is divided into a low gain combination, a medium gain combination and a high gain combination according to the size of the gain after the combination. In this embodiment, the combination is divided into three categories according to the gain percentage: low gain (30%-50%], medium gain (50%-70%], and high gain (70%-+∞).
[0064] Step S4, eliminating the combinations whose correlations between derived combinations are greater than a preset correlation threshold, and the combinations whose correlations with the original credit basic variables are greater than a preset correlation threshold, and selecting the target credit variable combination based on the post-combination gain and correlation.
[0065] It is understandable that in the process of combination mining, multiple highly correlated combination variables may be generated. Although these variables look different, they actually provide similar information, which will lead to increased redundancy and complexity of the model, and may also cause multicollinearity problems, affecting the stability and predictive ability of the model.
[0066] To address this problem, this embodiment eliminates highly correlated credit variable combinations to ensure that each combination is independent, which helps to improve the stability and predictive ability of the model, and can also reduce redundant information in the model, making the model more concise and efficient.
[0067] Specifically, in this embodiment, the correlation threshold is set to cor=0.7. cor (correlation coefficient) is the correlation coefficient. Credit variable combinations with correlation cor≥0.7 between derived combinations are eliminated, and credit variable combinations with correlation cor≥0.7 with the original credit basic variables are eliminated.
[0068] Furthermore, high-gain combinations whose correlations meet the requirements are selected as target credit variable combinations to obtain independent high-value combinations that can be used in production.
[0069] In this embodiment, the final independent high-value credit variable combination method is written in the form of a model interface, which is convenient for the downstream risk control system to call the combination result through the post request interface, and can realize the rapid application and promotion of credit variable combination mining technology.
[0070] The present invention combines and derives basic credit variables through machine learning algorithms and mathematical methods, generates new credit variable combinations, and calculates the value of these combinations, thereby deeply mining the potential value of credit data. The present invention optimizes the credit data's ability to distinguish good from bad customers by calculating the KS value of the credit variable combination and evaluating the gain after the combination.
[0071] Figure 2 This is a structural diagram of the credit variable combination mining system provided by the present invention, referring to Figure 2 , the credit variable combination mining system 200 includes:
[0072] The acquisition module 201 is used to acquire the historical credit data of the customer and extract the basic credit variables;
[0073] The combination derivation module 202 is used to combine and derive the basic credit variables using a machine learning algorithm. For each newly generated credit variable combination, the value of the credit variable combination is calculated through a pre-built calculation framework, and the credit variable combination whose value meets the standard is screened;
[0074] The gain calculation module 203 is used to calculate the KS value of the credit investigation basic variable and the credit investigation variable combination composed of it, and calculate the gain after the combination; wherein the KS value is used to evaluate the ability of the credit investigation basic variable to distinguish between good and bad customers;
[0075] The combination selection module 204 is used to eliminate the combinations whose correlation between the derived combinations is greater than the preset correlation threshold, and the combinations whose correlation with the original credit basic variables is greater than the preset correlation threshold, and select the target credit variable combination based on the post-combination gain and correlation.
[0076] The credit variable combination mining system provided by the present invention executes the credit variable combination mining method provided by the aforementioned embodiments through the above modules. The credit variable combination mining method has been described in detail in the above embodiments, and will not be repeated in this embodiment.
[0077] Figure 3 The structural block diagram of the electronic device provided by the present invention is as follows: Figure 3As shown, the present invention also provides an electronic device, and the electronic device 300 can be a computing device such as a mobile terminal, a desktop computer, a notebook, a palm computer, and a server. The electronic device 300 includes a processor 301 and a memory 302, wherein the memory 302 stores a credit variable combination mining program 303.
[0078] In some embodiments, the memory 302 may be an internal storage unit of a computer device, such as a hard disk or memory of the computer device. In other embodiments, the memory 302 may also be an external storage device of a computer device, such as a plug-in hard disk equipped on the computer device, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. Furthermore, the memory 302 may also include both an internal storage unit of a computer device and an external storage device. The memory 302 is used to store application software and various types of data installed on the computer device, such as program codes for installing the computer device. The memory 302 may also be used to temporarily store data that has been output or is to be output. In one embodiment, when the credit variable combination mining program 303 is executed by the processor 301, the following steps are implemented:
[0079] Obtain the customer's historical credit data and extract basic credit variables;
[0080] Use machine learning algorithms to combine and derive basic credit variables. For each newly generated credit variable combination, calculate the value of the credit variable combination through a pre-built calculation framework, and screen the credit variable combinations that meet the standards;
[0081] Calculate the KS value of the credit investigation basic variables and the credit investigation variable combination they constitute, and calculate the gain after the combination; the KS value is used to evaluate the ability of the credit investigation basic variables to distinguish between good and bad customers;
[0082] Eliminate combinations whose correlations between derived combinations are greater than the preset correlation threshold, and combinations whose correlations with the original credit basic variables are greater than the preset correlation threshold, and select the target credit variable combination based on the post-combination gain and correlation.
[0083] In some embodiments, the processor 301 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run program codes or process data stored in the memory 302, such as executing a credit variable combination mining program.
[0084] This embodiment further provides a computer-readable storage medium on which a credit variable combination mining program is stored. When the credit variable combination mining program is executed by a processor, the following steps are implemented:
[0085] Obtain the customer's historical credit data and extract basic credit variables;
[0086] Use machine learning algorithms to combine and derive basic credit variables. For each newly generated credit variable combination, calculate the value of the credit variable combination through a pre-built calculation framework, and screen the credit variable combinations that meet the standards;
[0087] Calculate the KS value of the credit basic variables and the credit variable combination they constitute, and calculate the gain after the combination; the KS value is used to evaluate the ability of the credit basic variables to distinguish between good and bad customers;
[0088] Eliminate combinations whose correlations between derived combinations are greater than the preset correlation threshold, and combinations whose correlations with the original credit basic variables are greater than the preset correlation threshold, and select the target credit variable combination based on the post-combination gain and correlation.
[0089] The above-described embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the attached claims.
[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A credit variable combination mining method, characterized in that: include: Obtain the customer's historical credit data and extract basic credit variables; Use machine learning algorithms to combine and derive basic credit variables. For each newly generated credit variable combination, calculate the value of the credit variable combination through a pre-built calculation framework, and screen the credit variable combinations that meet the standards; Calculate the KS value of the credit basic variables and the credit variable combination they constitute, and calculate the gain after the combination; the KS value is used to evaluate the ability of the credit basic variables to distinguish between good and bad customers; Eliminate combinations whose correlations between derived combinations are greater than the preset correlation threshold, and combinations whose correlations with the original credit basic variables are greater than the preset correlation threshold, and select the target credit variable combination based on the post-combination gain and correlation.
2. The credit variable combination mining method according to claim 1, characterized in that: The step of obtaining the historical credit data of the customer and extracting the basic credit variables includes: Extract historical credit data of multiple customers across platforms; Determine the criteria for dividing good and bad customers, and divide customers into good customer samples and bad customer samples based on their historical credit data; Extract basic information units used to describe personal credit status from the customer's historical credit data as basic credit variables, and integrate the basic credit variables into a wide table.
3. The credit variable combination mining method according to claim 1, characterized in that: The method of using a machine learning algorithm to combine and derive basic credit variables includes: Use machine learning algorithms or mathematical methods to combine and derive basic credit variables; Among them, the machine learning algorithms include K-means clustering, PCA principal component analysis and xgboost extreme gradient boosting; the mathematical methods include addition combination, subtraction combination, multiplication combination, division combination, square sum combination and trigonometric function combination.
4. The credit variable combination mining method according to claim 1, characterized in that: The method of combining and deriving the basic credit variables using a machine learning algorithm also includes: Use the for loop to traverse and combine the basic credit variables to generate a new credit variable combination.
5. The credit variable combination mining method according to claim 1, characterized in that: Before calculating the value of the credit variable combination through a pre-built calculation framework, the method further includes: Build IV calculation framework and KS calculation framework; Among them, the IV value calculated by the IV calculation framework is used to evaluate the impact of the basic credit variables on good and bad customers; the IV value and KS value are used to jointly judge the value of the basic credit variables.
6. The credit variable combination mining method according to claim 1, characterized in that: After calculating the KS value of the credit investigation basic variable and the credit investigation variable combination composed of the basic variable and calculating the combined gain, the method further includes: According to the size of the combined gain, the credit variable combinations are divided into low-gain combinations, medium-gain combinations and high-gain combinations; Accordingly, the target credit variable combination is selected based on the combined gain and correlation, including: Select a high-gain combination whose correlation meets the requirements as the target credit variable combination.
7. A credit variable combination mining system applied to the credit variable combination mining method according to any one of claims 1 to 6, characterized in that: include: The acquisition module is used to obtain the customer's historical credit data and extract the basic credit variables; The combination derivation module is used to combine and derive basic credit variables using machine learning algorithms. For each newly generated credit variable combination, the value of the credit variable combination is calculated through a pre-built calculation framework, and the credit variable combination whose value meets the standard is screened; The gain calculation module is used to calculate the KS value of the credit investigation basic variables and the credit investigation variable combination they constitute, and calculate the gain after the combination; wherein the KS value is used to evaluate the ability of the credit investigation basic variables to distinguish between good and bad customers; The combination selection module is used to eliminate the combinations whose correlation between derived combinations is greater than the preset correlation threshold, and the combinations whose correlation with the original credit basic variables is greater than the preset correlation threshold, and select the target credit variable combination based on the post-combination gain and correlation.
8. An electronic device, It is characterized in that comprising a memory and a processor, wherein: The memory is used to store programs; The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the credit variable combination mining method described in any one of claims 1 to 6 above.
9. A computer-readable storage medium, characterized in that: Used to store computer-readable programs or instructions, which, when executed by a processor, can implement the steps in the credit variable combination mining method described in any one of claims 1 to 6 above.