A method to improve the stability of risk control models

Abstract

The present invention relates to the technical field of credit risk assessment, and specifically to a method for improving the stability of a risk control model. The method comprises identifying and analyzing an enterprise's preset risk control model to determine a credit assessment map of the preset risk control model; comparing and analyzing the credit assessment maps of the enterprise's existing preset risk control models based on the preset stability map to find the specific reasons for the lack of stability of the preset risk control model currently used by the enterprise; and then, based on the specific reasons, specifically determining available algorithms and data processing processes to improve the stability of the preset risk control model and provide customers with reasonable assessment limits, thereby solving the problem of low stability of customer group data analyzed by existing risk control models when the economic environment changes.

Description

Technical Field

[0001] The present invention relates to the technical field of credit risk assessment, and specifically to a method for improving the stability of a risk control model. Background Art

[0002] In the field of credit risk assessment, the gradient boosting algorithm is widely used due to its excellent performance. In principle, this serial algorithm continuously corrects prediction biases on training samples to learn the complex nonlinear relationships between features and labels. Compared to traditional logistic regression and random forest algorithms, this inherently carries a higher risk of overfitting. In practical applications, changes in the economic environment can significantly alter the distribution of customer data, impacting the stability of these risk control models. Improving the stability of risk control models has always been a key concern for financial institutions to ensure smooth business operations.

[0003] Before training a machine learning model, the user must customize the evaluation function and hyperparameters. Ideal evaluation functions and hyperparameters can significantly improve the stability of the risk control model. Currently, mainstream risk control model evaluation metrics in the industry are typically KS, AUC, or the top-tier bad customer detection rate. KS measures the maximum absolute difference between the cumulative distribution of good and bad customers predicted by the model, reflecting the risk control model's ability to distinguish between good and bad customers. However, this metric focuses solely on the point where the risk control model's ability to distinguish is maximized, ignoring the overall distribution. Therefore, it provides an incomplete assessment of the risk control model's overall performance. AUC, by calculating the area under the receiver operating characteristic (ROC) curve, comprehensively reflects the performance of the risk control model under all possible thresholds. However, this metric is sensitive to extreme values ​​and, because it does not consider cost-effectiveness, does not directly reflect the risk control model's performance in real applications. The top-tier bad customer detection rate measures the proportion of actual bad customers among the top 10% or 20% of customers predicted by the model. It reflects the risk control model's ability to identify high-risk customers. However, this metric overly relies on the risk control model's predictions for high-risk customers and may overlook the accuracy of predictions for other customer groups.

[0004] Mainstream hyperparameter optimization methods currently used in the industry include grid search and random search. Grid search is an exhaustive search strategy that ensures the optimal hyperparameter combination within a specified range is found. However, when the range of hyperparameter values ​​is large, its computational cost is too high, making it difficult to apply in real-world scenarios. Unlike grid search, random search randomly selects hyperparameter values ​​from a predefined hyperparameter distribution for evaluation. This strategy can more efficiently utilize computing resources, but its results are somewhat random and do not consider historical evaluation information.

[0005] Therefore, the present invention provides a method for improving the stability of a risk control model to solve the above problems. Summary of the Invention

[0006] In view of the above situation, in order to overcome the defects of the existing technology, the present invention provides a method for improving the stability of a risk control model to solve the problem of low stability of the existing risk control model in analyzing customer group data when the economic environment changes.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is:

[0008] A method for improving the stability of a risk control model includes: identifying and analyzing a preset risk control model of an enterprise to determine a credit assessment map of the preset risk control model; obtaining the enterprise's regulatory demand information, and based on customer group information and regulatory demand information, analyzing the credit assessment map using a preset stability map to generate an adjustment strategy;

[0009] Based on the adjustment strategy, adjust the preset risk control model to obtain the initial risk control model;

[0010] Obtain dynamic credit assessment data for customer groups and supplementary data related to adjustment strategies;

[0011] Based on preset dimensional indicators, a clustering algorithm is used to process the supplementary data and dynamic credit assessment data to establish a multi-dimensional indicator data set; and the multi-dimensional indicator data set is divided into a training set, a test set and multiple validation sets; the training set, the test set and multiple validation sets are used to process the initial risk control model to obtain the target risk control model.

[0012] Preferably, the identification and analysis of the preset risk control model of the enterprise to determine the credit assessment map of the preset risk control model includes: obtaining the construction record information and current program information of the preset risk control model; determining the construction framework of the preset risk control model based on the construction record information; determining the processing program corresponding to the framework based on the construction framework and the remark information in the program information; analyzing the processing program to determine the data processing algorithm and the data processing flow; and determining the credit assessment map based on the construction framework, the data processing algorithm and the data processing flow.

[0013] Preferably, the credit assessment map is analyzed based on the customer group information and the regulation demand information using a preset stability map to generate an adjustment strategy, including: extracting features from the customer group information and the regulation demand information to obtain key information; comparing the preset stability map and the credit assessment map in a framework-to-data processing flow to obtain assessment missing features that are lacking in the credit assessment map; adjusting the assessment missing features according to the key information to obtain initial adjustment features; and obtaining the corresponding construction framework and / or data processing algorithm and / or data processing flow from the preset stability map according to the initial adjustment features to generate an adjustment strategy.

[0014] Preferably, the preset risk control model is adjusted based on the adjustment strategy to obtain the initial risk control model, including: generating a data processing program according to the data processing algorithm and / or data processing flow in the adjustment strategy; generating an associated structure document of the data processing program according to the relationship between the program documents in the preset risk control model; adding the associated structure program with the data processing program to the corresponding position of the preset risk control model to obtain the initial risk control model.

[0015] Preferably, the obtaining of supplementary data related to the adjustment strategy includes: determining the supplementary data type and the supplementary data name according to the program corresponding to the adjustment strategy; analyzing the supplementary data type and the supplementary data name to determine the source location of the supplementary data; and collecting information corresponding to the supplementary data type and the supplementary data name according to a preset data collection method corresponding to the source location of the supplementary data to obtain the supplementary data.

[0016] Preferably, the method uses a clustering algorithm based on preset dimensional indicators to process the supplementary data and dynamic credit assessment data to establish a multi-dimensional indicator data set, including: using a hierarchical clustering algorithm and a K-Means clustering algorithm to perform cluster analysis on the supplementary data and dynamic credit assessment data to obtain a first cluster type, and obtaining a dimensional data set of the same type according to the first cluster type and each preset cluster type of the preset dimensional indicators; and using the STING algorithm to remove or add data in the first cluster type that is different from the preset cluster type to obtain a multi-dimensional indicator data set.

[0017] Preferably, the multidimensional indicator data set is divided into a training set, a test set and multiple validation sets according to preset classification rules, including: selecting a preset number of data sets with the same time span and similar data distribution from the multidimensional indicator data set, and dividing them into training sets and test sets according to a preset ratio; selecting data with different performance periods and different overdue degrees from the multidimensional indicator data set as a validation set.

[0018] Preferably, the method of processing the initial risk control model using multiple validation sets to obtain the target risk control model includes: optimizing the hyperparameters of the initial risk control model using a greedy strategy, a Bayesian optimization algorithm, and a grid search method.

[0019] The beneficial effects of the present invention are:

[0020] 1. The present invention compares and analyzes the credit assessment maps of the enterprise's existing preset risk control models based on the preset stability map, and finds the specific reasons for the lack of stability of the preset risk control models currently used by the enterprise; then, based on the specific reasons, the available algorithms and data processing processes are targeted to adjust the preset risk control models currently used by the enterprise, and obtain a target risk control model with enhanced stability, thereby solving the problem of low stability of the enterprise's existing risk control models, and helping to solve the problem of low stability of customer group data analyzed by existing risk control models when the economic environment changes.

[0021] 2. Based on customer group information and regulatory demand information, the present invention can use a preset stability map to analyze the credit assessment map and generate an adjustment strategy; by introducing the enterprise's regulatory demand information, the preset risk control model can be adjusted in a targeted manner so that the final target risk control model fits the needs of the enterprise.

[0022] 3. When optimizing the hyperparameters of the initial risk control model after testing using the validation set, a greedy strategy was used, employing a Bayesian optimization algorithm and a grid search method. This combination of greedy strategies, Bayesian optimization algorithms, and grid search methods effectively optimizes the hyperparameters of the initial risk control model, helping to improve the model's stability and generalization capabilities. This, in turn, helps address the issue of low stability in analyzing customer group data with existing risk control models when the economic environment changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 The figure is a schematic flow chart of a method for improving the stability of a risk control model according to the present invention. DETAILED DESCRIPTION

[0024] The following will refer to the attached Figure 1 The embodiments of the present invention are described in detail. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0025] A method for improving the stability of risk control models, as shown in the attached Figure 1 As shown, the following steps are included:

[0026] Step S11: Identify and analyze the enterprise's preset risk control model to determine the credit assessment map of the preset risk control model.

[0027] Step S12: Obtain the enterprise's regulation and control demand information, and analyze the credit assessment map using a preset stability map based on the customer group information and the regulation and control demand information to generate an adjustment strategy.

[0028] The preset stability map includes the data types used to assess user creditworthiness, data acquisition methods, the analytical process for assessing user creditworthiness, and the algorithms and processing steps used in each process step. The preset stability map is presented and analyzed in a layered, generalized to detailed manner, allowing for rapid comparison and analysis between the credit assessment map based on the preset risk control model and the preset stability map, quickly identifying the differences between the two.

[0029] Similarly, the structure and content of the credit assessment map are similar to those of the preset stability map.

[0030] Step S13: Based on the adjustment strategy, the preset risk control model is adjusted to obtain an initial risk control model.

[0031] Step S14: Acquire dynamic credit assessment data of customer groups and supplementary data related to adjustment strategies.

[0032] Among them, supplementary data refers to data that has not been analyzed in the company's preset risk control model, including but not limited to industry dynamics reports, government support policy information, etc.

[0033] Step S15: Based on the preset dimensional indicators, a clustering algorithm is used to process the supplementary data and dynamic credit assessment data to establish a multi-dimensional indicator data set; and the multi-dimensional indicator data set is divided into a training set, a test set, and multiple validation sets according to the preset classification rules; the initial risk control model is processed using the training set, the test set, and the multiple validation sets to obtain the target risk control model.

[0034] Among them, the preset dimension indicators include but are not limited to user information dimension indicators, financial dimension indicators, credit history dimension indicators, user behavior preference dimension indicators, social responsibility dimension indicators, risk characteristic dimension indicators, and development prospect dimension indicators.

[0035] Through the setting method of this embodiment, the present invention can compare and analyze the credit assessment map of the enterprise's existing preset risk control model based on the preset stability map, and find the specific reasons for the lack of stability of the preset risk control model currently used by the enterprise; and then determine the available algorithms and data processing processes based on the specific reasons, so as to improve the stability of the preset risk control model and give customers a reasonable assessment limit.

[0036] Furthermore, based on the preset stability map, a comparative analysis is conducted on the credit assessment maps of the company's existing preset risk control models to find the specific reasons for the lack of stability of the preset risk control models currently used by the company; then, based on the specific reasons, the available algorithms and data processing processes are targeted to adjust the preset risk control models currently used by the company, and a target risk control model with enhanced stability is obtained, thereby solving the problem of low stability of the company's existing risk control models and helping to solve the problem of low stability of customer group data analyzed by existing risk control models when the economic environment changes.

[0037] In one embodiment of the present invention, an enterprise's preset risk control model is identified and analyzed to determine a credit assessment map of the preset risk control model, including: obtaining the construction record information and current program information of the preset risk control model; determining the construction framework of the preset risk control model based on the construction record information; determining the processing program corresponding to the framework based on the construction framework and the remark information in the program information; analyzing the processing program to determine the data processing algorithm and the data processing flow; and determining the credit assessment map based on the construction framework, the data processing algorithm and the data processing flow.

[0038] The build record information for the preset risk control model includes, but is not limited to, update information for each version of the preset risk control model, the initial build plan, the build algorithm, and debugging information. The current program information for the preset risk control model includes the front-end program information, back-end program information, database program information, and test program information involved in the entire risk control model. The data processing flow includes the selection logic, judgment logic, and loop logic for data processing within the program.

[0039] When determining the construction framework of the preset risk control model based on the construction record information, a text processing model is used to perform comprehensive recognition and analysis of the construction record information to obtain the overall construction framework of the preset risk control model. The text processing model can be a bag-of-words model, a text feature extraction model based on a neural network, etc., which can perform text feature extraction to obtain a model of the construction framework.

[0040] Furthermore, when part of the program information does not contain remark information, the package import information and specific program in the part of the program will be analyzed to determine which part of the information in the construction framework the program belongs to; and the content of the specific program will be analyzed to determine the data processing algorithm and data processing flow of the part of the program.

[0041] Analyzing the processing program to determine the data processing algorithm and data processing flow can be done by: using one or more of the abstract syntax tree (AST) analysis method, the control flow graph (CFG) analysis method, the data flow analysis method, etc. to analyze the processing program to determine the specific data processing algorithm and data processing flow of the processing program, thereby being able to comprehensively and accurately construct a credit assessment map of the preset risk control model.

[0042] Through the setting method of this embodiment, the present invention can identify and analyze the enterprise's preset risk control model to determine the credit assessment map of the preset risk control model, so that the determined credit assessment map is accurate and comprehensive, and thus the assessment missing features obtained by subsequent comparative analysis are more accurate and reliable.

[0043] In one embodiment of the present invention, based on customer group information and regulation demand information, a preset stability map is used to analyze the credit assessment map to generate an adjustment strategy, including: extracting features from the customer group information and regulation demand information to obtain key information; comparing the preset stability map and the credit assessment map in a framework-to-data processing flow to obtain assessment missing features that are lacking in the credit assessment map; adjusting the assessment missing features according to the key information to obtain initial adjustment features; and obtaining the corresponding construction framework and / or data processing algorithm and / or data processing flow from the preset stability map according to the initial adjustment features to generate an adjustment strategy.

[0044] Among them, the methods for obtaining regulatory demand information include text acquisition, voice acquisition, or video acquisition; the method for obtaining customer group information is obtained by analyzing customer information stored in the database. When extracting features from customer group information and regulatory demand information to obtain key information, a text feature extraction model is used to extract features from these customer group information and regulatory demand information, and determine the target analysis features of this information in order to determine the focus of the adjustment of the preset risk control model required by the enterprise; for example, the data processing results of the adjusted preset risk control model are more reasonable, the data analysis capabilities of the preset risk control model are more stable, the data analysis speed of the preset risk control model is faster, and the program protection of the preset risk control model is more secure.

[0045] When comparing the preset stability map and the credit assessment map in the manner of framework to data processing flow, a comparative analysis will be conducted in the order from the type in the framework to the specific implementation method, so as to comprehensively identify the framework types and specific implementation methods that are missing in the credit assessment map.

[0046] Through the above-described embodiments, the present invention can analyze the credit assessment map using a preset stability map based on customer group information and regulatory demand information, generating an adjustment strategy corresponding to the credit assessment map that comprehensively covers and meets the needs of the enterprise. This adjustment strategy can then make the preset risk control model, adjusted using the adjustment strategy, more aligned with the actual needs of the enterprise, resulting in more reasonable credit assessment results and greater assessment stability for the preset risk control model. Furthermore, by incorporating the enterprise's regulatory demand information, the preset risk control model can be adjusted in a targeted manner, resulting in a target risk control model that meets the enterprise's needs.

[0047] In one embodiment of the present invention, the preset risk control model is adjusted based on the adjustment strategy to obtain an initial risk control model, including: generating a data processing program according to the data processing algorithm and / or data processing process in the adjustment strategy; generating an associated structure document of the data processing program according to the relationship between the program documents in the preset risk control model; adding the associated structure program with the data processing program to the corresponding position of the preset risk control model to obtain the initial risk control model.

[0048] In this embodiment, according to the corresponding adjustment content in the adjustment strategy, the corresponding data processing program is found from the preset program corresponding library, and then, based on the generated associated structure document, the data processing program is added to the associated structure document to form an adjustment document corresponding to this part of the adjustment strategy; and then the adjustment document is added to the corresponding program project of the corresponding module in the preset risk control model.

[0049] After each adjustment document is added to the preset risk control model, debugging is performed based on the program document associated with the adjustment document, and calls between program documents are tested to determine whether there is any conflict between the debugging document and the existing program document.

[0050] Through the above-described implementation, the present invention can perform targeted processing on the preset risk control model based on the adjustment strategy, generate an adjustment document that fits the preset risk control model, and obtain an initial risk control model. Compared with the traditional manual inspection and adjustment method, through the configuration of this embodiment, the present invention can automatically generate an adjustment document corresponding to the preset risk control model based on the adjustment strategy, saving manual processing time, improving data processing speed, and avoiding interference from human factors.

[0051] In one embodiment of the present invention, obtaining supplementary data related to the adjustment strategy includes: determining the supplementary data type and the supplementary data name according to the program corresponding to the adjustment strategy; analyzing the supplementary data type and the supplementary data name to determine the source location of the supplementary data; and collecting information corresponding to the supplementary data type and the supplementary data name according to a preset data collection method corresponding to the source location of the supplementary data to obtain the supplementary data.

[0052] Among them, after obtaining the initial risk control model, in order to train and adjust the parameters of the initial risk control model, the newly added supplementary data types and supplementary data names in the adjustment strategy will be obtained first; then the source location of these data will be determined so that they can be divided into corresponding data collection methods; and these data will be obtained through the corresponding data collection methods.

[0053] For example, when additional support policy information is needed, web crawlers are used to retrieve relevant support policy information from official websites with high authenticity for the corresponding time period. When additional information on the current state of an industry is needed, web crawlers are used to obtain articles and videos with high authenticity within a preset time period. Natural language processing models are then used to analyze these articles and videos to obtain information on the current state of the industry. When additional personal information is needed, customers are asked to provide the relevant information, which is then stored in the corresponding personal information database.

[0054] Through the configuration of this embodiment, the present invention can accurately obtain the adjustment data corresponding to the adjustment strategy. After obtaining this supplementary data, the data can be processed and divided into a training set, a test set, and a validation set. Then, the initial risk control model is trained, the trained initial risk control model is tested, and the validation set is used for verification. After passing the verification, the final target risk control model that meets the enterprise's needs is obtained.

[0055] In one embodiment of the present invention, based on preset dimensional indicators, a clustering algorithm is used to process the supplementary data and dynamic credit assessment data to establish a multi-dimensional indicator data set, including: using a hierarchical clustering algorithm and a K-Means clustering algorithm to perform cluster analysis on the supplementary data and dynamic credit assessment data to obtain a first cluster type, and obtaining a dimensional data set of the same type according to the first cluster type and each preset cluster type of the preset dimensional indicators; using the STING algorithm to remove or add data in the first cluster type that is different from the preset cluster type to obtain a multi-dimensional indicator data set.

[0056] In this embodiment, the processing process of the clustering method using a combination of a hierarchical clustering algorithm and a K-Means clustering algorithm is as follows: the hierarchical clustering algorithm is used to perform preliminary clustering on the data. Since hierarchical clustering does not require pre-specified cluster numbers, it can flexibly handle data with different distributions. During the hierarchical clustering process, an appropriate agglomerative or divisive method is selected based on the data characteristics and clustering requirements to improve clustering accuracy. The cluster tree obtained by hierarchical clustering can intuitively observe the clustering structure of the data and determine the appropriate number of clusters K accordingly. This helps to avoid the problem of local optimal solutions caused by the random selection of initial cluster centers in the K-Means algorithm. Based on the determined number of clusters K, K representative cluster centers are selected from the results of the hierarchical clustering as the initial cluster centers of the K-Means algorithm. These centers can more accurately reflect the actual distribution of the data. The K-Means algorithm is executed using the selected initial cluster centers. Because the selection of initial cluster centers is more reasonable, the K-Means algorithm can converge to the global optimal solution more quickly, thereby obtaining the final first cluster type. By combining the hierarchical clustering algorithm and the K-Means clustering algorithm, the present invention can give full play to the advantages of both algorithms and improve the accuracy and stability of clustering.

[0057] When the STING algorithm is used to remove or add data in the first cluster type that is different from the preset cluster type to obtain a multi-dimensional indicator data set, for data points in the first cluster type that are different from the preset cluster type, STING will assign them to the most similar cluster based on the statistical information of the grid unit where the data points are located, or regard them as noise and remove them.

[0058] Through the above-mentioned approach, the present invention accurately and comprehensively classifies supplementary data and dynamic credit assessment data. By combining the hierarchical clustering algorithm with the K-Means clustering algorithm, the present invention leverages the advantages of both to improve clustering accuracy and stability. Furthermore, the STING algorithm is used to process data that does not conform to the preset clustering type, reducing data rejection and facilitating the rapid accumulation of training data for the initial risk control model.

[0059] In one embodiment of the present invention, a multidimensional indicator dataset is divided into a training set, a test set, and multiple validation sets according to preset classification rules, including: selecting a preset number of datasets with the same time span and similar data distribution from the multidimensional indicator dataset, and dividing them into training sets and test sets according to a preset ratio; selecting data with different performance periods and different overdue degrees from the multidimensional indicator dataset as validation sets.

[0060] After obtaining the preset amount of data for each type, the training set and test set are divided into a preset ratio of 7:3. During the process of processing the initial risk control model using the training set, test set, and multiple validation sets to obtain the target risk control model, the test set is used to evaluate the performance of the trained initial risk control model, that is, the training of the initial risk control model is evaluated by calculating indicators such as accuracy, recall rate, and F1 value. After passing the test set test, the validation set is used for cross-validation: the optimal hyperparameter combination is found through multiple training and validation, the final target risk control model is obtained, and the target risk control model is deployed in the actual environment.

[0061] During the hyperparameter tuning using the validation set, based on the greedy strategy, the Bayesian optimization algorithm and grid search method are used to optimize the hyperparameters of the initial risk control model; the existing evaluation functions and hyperparameters in the initial risk control model after the test are optimized to improve the stability of the risk control model. When optimizing the evaluation function: by adding different evaluation data sets for different time periods, selecting appropriate evaluation indicators, and then weightedly adjusting the evaluation function for hyperparameter optimization, the cross-time stability of the model is improved while meeting the business application requirements. When optimizing hyperparameters, the combined design method of Bayesian optimization and grid search is used to reduce the time for adjusting the parameters of the risk control model and improve the performance of the risk control model. In the above manner, the present invention can greatly shorten the time for model training and parameter adjustment and effectively improve the generalization and stability of the model on cross-period samples.

[0062] Among them, when optimizing the evaluation function, the underlying indicators in the evaluation function are selected according to the business application requirements. : When you need to improve the risk control model's ability to identify risks in all samples, select AUC; when you need to improve the risk control model's ability to distinguish between good and bad samples, select KS; when you need to improve the risk control model's ability to identify high-risk customers, select the top bad rate.

[0063] Among them, the evaluation function of hyperparameter optimization for:

[0064] ,

[0065] in, is an adjustable penalty coefficient, For the test indicators, are all the underlying indicators of the evaluation function, and the total number of underlying indicators is n. is the training indicator, The test indicator evaluation function is used to evaluate the degree of parameter optimization based on the data corresponding to each test indicator.

[0066] When optimizing hyperparameters, in a larger hyperparameter space The Bayesian optimization method is used to maximize the objective function, and a smaller hyperparameter space is constructed based on the hyperparameter results of each round of optimization. Finally In the algorithm, each hyperparameter is gradually optimized using grid search based on a greedy strategy.

[0067] Through the configuration of this embodiment, the present invention can optimize the existing evaluation functions and hyperparameters in the initial risk control model after the test, thereby improving the stability of the target risk control model.

[0068] Furthermore, based on the greedy strategy, the hyperparameters of the initial stability model are optimized by combining the Bayesian optimization algorithm with the grid search method. One execution method is:

[0069] Define the hyperparameter search space (including the name, type, and value range of each hyperparameter). Initialize the surrogate model (e.g., Gaussian process) for the Bayesian optimization algorithm. Use a greedy strategy to select the optimal hyperparameter combination for evaluation based on the hyperparameter performance predicted by the surrogate model. Update the surrogate model to reflect the new observations. Repeat these steps until the preset number of Bayesian optimization iterations is reached or a satisfactory hyperparameter combination is found. Define a finer grid around the optimal hyperparameter combination found by Bayesian optimization. Use a grid search method to try all possible hyperparameter combinations within the defined grid. Select the hyperparameter combination with the best performance as the final result. Evaluate the performance of the final hyperparameter combination. Based on the evaluation results, decide whether to accept the hyperparameter combination as the final configuration of the model.

[0070] Specifically, the process involves defining a hyperparameter search space, determining the hyperparameters to be optimized and their possible value ranges. For example, for a support vector machine model, these hyperparameters may include the regularization parameter C and the kernel function type. Initialize the Bayesian optimization algorithm, select an appropriate acquisition function (such as Expected Improvement), and set the initial state of the surrogate model. At each step, based on the surrogate model's predicted hyperparameter performance, select the optimal hyperparameter combination for actual evaluation. Evaluate the selected hyperparameter combination using a separate validation set to obtain actual performance metrics. Update the surrogate model using the new observations (hyperparameter combinations and their corresponding performance metrics) to improve its prediction accuracy. Repeat these steps until the preset number of iterations is reached or a satisfactory hyperparameter combination is found. Define a finer grid around the optimal hyperparameter combination found by Bayesian optimization. The granularity of the grid can be adjusted based on the specific problem and computing resources. Within the defined grid, try all possible hyperparameter combinations one by one, and record the performance metrics for each combination. Compare the performance metrics of all hyperparameter combinations within the grid, and select the one with the best performance as the final result.

[0071] The final selected hyperparameter combination is evaluated using an independent validation set to verify its generalization ability. If the evaluation results meet the requirements, the hyperparameter combination is used for the final deployment of the model; otherwise, further adjustments to the hyperparameter search space or optimization algorithm can be considered.

[0072] Through the above steps, combined with the greedy strategy, Bayesian optimization algorithm and grid search method, the hyperparameters of the initial risk control model can be effectively optimized to improve the performance and generalization ability of the model.

[0073] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0074] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0075] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fibers, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0076] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0077] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0078] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0079] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. A method for improving the stability of a risk control model, characterized in that: include: Identify and analyze the enterprise's preset risk control model to determine a credit assessment map for the preset risk control model, including: obtaining the construction record information and current program information of the preset risk control model; determining the construction framework of the preset risk control model based on the construction record information; determining the processing procedures corresponding to the framework based on the construction framework and the remarks in the program information; analyzing the processing procedures to determine the data processing algorithm and data processing flow; and determining a credit assessment map based on the construction framework, data processing algorithm, and data processing flow; Obtaining the enterprise's regulation and control demand information, and based on the customer group information and the regulation and control demand information, analyzing the credit assessment map using a preset stability map to generate an adjustment strategy, including: extracting features from the customer group information and the regulation and control demand information to obtain key information; comparing the preset stability map with the credit assessment map in a framework-to-data processing flow to obtain missing assessment features that are lacking in the credit assessment map; adjusting the missing assessment features based on the key information to obtain initial adjustment features; and obtaining a corresponding construction framework and / or data processing algorithm and / or data processing flow from the preset stability map based on the initial adjustment features to generate an adjustment strategy; Based on the adjustment strategy, the preset risk control model is adjusted to obtain an initial risk control model; the steps include: generating a data processing program according to the data processing algorithm and / or data processing flow in the adjustment strategy; generating an associated structure document of the data processing program according to the relationship between program documents in the preset risk control model; and adding the associated structure program with the data processing program added to the corresponding position of the preset risk control model to obtain the initial risk control model; Obtain dynamic credit assessment data for customer groups and supplementary data related to adjustment strategies; Based on preset dimensional indicators, a clustering algorithm is used to process the supplementary data and dynamic credit assessment data to establish a multi-dimensional indicator dataset. The multi-dimensional indicator dataset is then divided into a training set, a test set, and multiple validation sets according to preset classification rules. The initial risk control model is processed using the training set, the test set, and the multiple validation sets to obtain the target risk control model. The method of dividing the multidimensional indicator data set into a training set, a test set and multiple validation sets according to preset classification rules includes: selecting a preset number of data sets with the same time span and similar data distribution from the multidimensional indicator data set, and dividing them into training sets and test sets according to a preset ratio; selecting data with different performance periods and different overdue degrees from the multidimensional indicator data set as validation sets.

2. The method for improving the stability of the risk control model according to claim 1, characterized in that: The obtaining of supplementary data related to the adjustment strategy includes: determining the supplementary data type and the supplementary data name according to the program corresponding to the adjustment strategy; analyzing the supplementary data type and the supplementary data name to determine the source location of the supplementary data; and collecting information corresponding to the supplementary data type and the supplementary data name according to a preset data collection method corresponding to the source location of the supplementary data to obtain the supplementary data.

3. The method for improving the stability of the risk control model according to claim 1, characterized in that: The method of processing the supplementary data and dynamic credit assessment data based on the preset dimensional indicators using a clustering algorithm to establish a multi-dimensional indicator data set includes: using a hierarchical clustering algorithm and a K-Means clustering algorithm to perform cluster analysis on the supplementary data and dynamic credit assessment data to obtain a first cluster type, and obtaining a dimensional data set of the same type based on the first cluster type and each preset cluster type of the preset dimensional indicators; and using the STING algorithm to remove or add data in the first cluster type that is different from the preset cluster type to obtain a multi-dimensional indicator data set.

4. The method for improving the stability of a risk control model according to claim 1, characterized in that: The method of processing the initial risk control model using multiple validation sets to obtain the target risk control model includes: optimizing the hyperparameters of the initial risk control model based on a greedy strategy using a Bayesian optimization algorithm and a grid search method.

Images