Credit overdue risk prediction method and device

Abstract

The embodiment of the present application provides a method and device for predicting credit overdue risk, which can be used in the field of artificial intelligence technology. The method includes: generating a corresponding target credit feature vector based on the credit data of the target user; inputting the target credit feature vector into a preset credit overdue risk prediction model, and determining the credit overdue risk prediction result of the target user based on the output of the credit overdue risk prediction model; the credit overdue risk prediction model is pre-trained based on balanced historical credit data, and the balanced historical credit data is data obtained after data balancing processing of pre-acquired unbalanced historical data. The present application can improve the distribution uniformity of the training data used for training the credit overdue risk prediction model, can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, and can effectively improve the accuracy and effectiveness of credit overdue risk prediction.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, in particular to the field of artificial intelligence technology, and specifically to a method and device for predicting credit overdue risk. Background Art

[0002] In the credit business, the qualifications and repayment ability of financial users directly determine the size of the financial risks that banks and other financial institutions have to bear. Since the credit reporting system in some areas is not yet perfect, when financial institutions assess the credit rating of financial users, in addition to referring to personal credit reports, they also need to rely on the basic information and asset information provided by the financial users themselves. This assessment method not only leads to some financial users reporting false assets or using borrowing to pass the review, but also requires professional data reviewers to analyze and judge, which is time-consuming and labor-intensive and has a low accuracy rate, which will lead to loan arrears or credit card defaults.

[0003] At present, existing technologies can predict credit defaults to improve the automation of credit overdue risk warning, such as based on statistics, operations research and artificial intelligence methods. However, although these relatively automated methods have shown good credit risk prediction capabilities, they ignore the defects of the data set to be analyzed, that is, the data in actual credit risk assessment is of various types and unstructured, especially the problem of data category imbalance. For example: convert the credit customer rating assessment problem into a binary classification problem. Assuming that there are 100 credit customers, 98 of them have good credit ratings and 2 have poor credit ratings, even if the learning algorithm we use always predicts the customer to be evaluated as having a good credit rating, the classification accuracy can reach 98%, that is, the majority of the data directly affects the performance of the classification algorithm. In this way, the learning algorithm trained by the unbalanced data set is of no value and meaning, and cannot verify the effectiveness of the entire risk prediction model.

[0004] Therefore, when studying the actual credit overdue risk prediction problem, how to effectively solve the imbalance problem of training data is crucial to improving the accuracy and reliability of credit risk prediction. Summary of the invention

[0005] In response to the problems in the prior art, the present application provides a method and device for predicting credit overdue risk, which can improve the distribution uniformity of training data used for training a credit overdue risk prediction model, and can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, thereby effectively improving the accuracy and effectiveness of credit overdue risk prediction.

[0006] In order to solve the above technical problems, this application provides the following technical solutions:

[0007] In a first aspect, the present application provides a method for predicting credit overdue risk, comprising:

[0008] Generate a corresponding target credit feature vector based on the credit data of the target user;

[0009] Inputting the target credit feature vector into a preset credit overdue risk prediction model, and determining a credit overdue risk prediction result of the target user based on an output of the credit overdue risk prediction model;

[0010] The credit overdue risk prediction model is pre-trained based on balanced historical credit data, and the balanced historical credit data is data obtained after data balancing processing is performed on pre-acquired unbalanced historical data.

[0011] Furthermore, before inputting the target credit feature vector into a preset credit overdue risk prediction model, the method further includes:

[0012] Acquire historical credit data and labels corresponding to each of a plurality of historical users, wherein the labels include: a first label for indicating that the corresponding historical user has a credit overdue risk and a second label for indicating that the corresponding historical user does not have a credit overdue risk;

[0013] Generate a corresponding first data set according to the historical credit data and labels corresponding to each of the historical users, and divide the first data set into a training set and a test set;

[0014] If the data in the training set is unbalanced historical data, performing data balancing processing on the training set so as to convert the data in the training set into balanced historical credit data and form a second data set;

[0015] Applying the second data set to obtain a corresponding credit feature vector;

[0016] The credit overdue risk prediction model is obtained based on the credit feature vector training.

[0017] Furthermore, before dividing the first data set into a training set and a test set, the method further includes:

[0018] The historical credit data corresponding to each of the historical users is preprocessed, wherein the preprocessing includes at least one of: completing missing data and deleting abnormal and invalid data.

[0019] Furthermore, if the data in the training set is unbalanced historical data, before performing data balancing processing on the training set so as to convert the data in the training set into balanced historical credit data and form a second data set, the method further includes:

[0020] If the ratio of the total number of historical users corresponding to the first label in the training set to the total number of historical users corresponding to the second label is greater than a first threshold or less than a second threshold, it is determined that the data in the training set is unbalanced historical data, wherein the first threshold is greater than 1 and the second threshold is greater than 0 and less than 1.

[0021] Furthermore, performing data balancing processing on the training set so as to convert the data in the training set into balanced historical credit data and form a second data set includes:

[0022] Determining minority class data and majority class data in the training set according to the total number of historical users corresponding to the first label and the total number of historical users corresponding to the second label in the training set;

[0023] Applying a preset data balancing method to perform K nearest neighbor interpolation processing on the minority class data, and performing K mean clustering processing on the majority class data, so as to convert the data in the training set into balanced historical credit data and form a corresponding second data set;

[0024] Wherein, the data balancing method includes: a preset SMOTE algorithm.

[0025] Furthermore, the applying the second data set to obtain a corresponding credit feature vector includes:

[0026] The second data set is subjected to data dimension reduction and feature screening processing to obtain credit feature vectors corresponding to each of the historical users and the virtual users obtained through data balancing processing.

[0027] Furthermore, the credit overdue risk prediction model is obtained based on the credit feature vector training, including:

[0028] Applying a preset random forest algorithm to obtain an initial prediction model based on the credit feature vector training;

[0029] The initial prediction model is evaluated for effectiveness according to the evaluation index corresponding to the test set, and the initial prediction model is adjusted based on the corresponding effect evaluation result to obtain the corresponding credit overdue risk prediction model.

[0030] In a second aspect, the present application provides a credit overdue risk prediction device, comprising:

[0031] A vector generation module, used to generate a corresponding target credit feature vector based on the credit data of the target user;

[0032] A model prediction module, used for inputting the target credit feature vector into a preset credit overdue risk prediction model, and determining the credit overdue risk prediction result of the target user based on the output of the credit overdue risk prediction model;

[0033] The credit overdue risk prediction model is pre-trained based on balanced historical credit data, and the balanced historical credit data is data obtained after data balancing processing is performed on pre-acquired unbalanced historical data.

[0034] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the credit overdue risk prediction method is implemented when the processor executes the program.

[0035] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which implements the credit overdue risk prediction method when executed by a processor.

[0036] It can be seen from the above technical scheme that the present application provides a method and device for predicting credit overdue risk. The method generates a corresponding target credit feature vector based on the credit data of the target user; inputs the target credit feature vector into a preset credit overdue risk prediction model, and determines the credit overdue risk prediction result of the target user based on the output of the credit overdue risk prediction model; the credit overdue risk prediction model is pre-trained based on balanced historical credit data, and the balanced historical credit data is data obtained after data balancing processing of pre-acquired unbalanced historical data. Data balancing processing is performed on the pre-acquired unbalanced historical data, which can effectively improve the distribution uniformity of the training data used for training the credit overdue risk prediction model, and can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, thereby effectively improving the accuracy and effectiveness of credit overdue risk prediction, and providing banks and other financial institutions with a more accurate and reliable judgment basis to carry out early warning processing or risk control processing of credit overdue risks, thereby effectively reducing the credit overdue risk of users and improving the user experience of financial institutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1It is a first flow chart of the credit overdue risk prediction method in the embodiment of the present application.

[0039] Figure 2 This is a second flow chart of the credit overdue risk prediction method in the embodiment of the present application.

[0040] Figure 3 It is a third flow chart of the credit overdue risk prediction method in the embodiment of the present application.

[0041] Figure 4 This is a fourth flow chart of the credit overdue risk prediction method in the embodiment of the present application.

[0042] Figure 5 It is a flowchart of step 030 in the credit overdue risk prediction method in the embodiment of the present application.

[0043] Figure 6 It is a flowchart of step 050 in the credit overdue risk prediction method in the embodiment of the present application.

[0044] Figure 7 It is a first structural schematic diagram of the credit overdue risk prediction device in the embodiment of the present application.

[0045] Figure 8 It is a second structural schematic diagram of the credit overdue risk prediction device in the embodiment of the present application.

[0046] Fig. 9 It is a logical flow chart of the credit overdue risk prediction method in the application example of this application.

[0047] Fig.10 It is a logical flow chart of the KF-SMOTE algorithm in the application example of this application.

[0048] Fig.11 It is a schematic diagram of the structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of this application.

[0050] It should be noted that the credit overdue risk prediction method and device disclosed in the present application can be used in the field of artificial intelligence technology, and can also be used in any field outside the field of artificial intelligence technology. The application field of the credit overdue risk prediction method and device disclosed in the present application is not limited.

[0051] Although the existing credit behavior default prediction methods have shown that they have good credit risk prediction capabilities, they ignore the defects of the data set to be analyzed, that is, the data in the actual credit risk assessment are of various types and unstructured, especially there is an imbalance problem in data categories. The present application provides a credit overdue risk prediction method, a credit overdue risk prediction device, an electronic device and a computer-readable storage medium, which generates a corresponding target credit feature vector based on the credit data of the target user; inputs the target credit feature vector into a preset credit overdue risk prediction model, and determines the credit overdue risk prediction result of the target user based on the output of the credit overdue risk prediction model; wherein the credit overdue risk prediction model predicts The balanced historical credit data is trained based on balanced historical credit data, and the balanced historical credit data is the data obtained after data balancing processing is performed on the pre-acquired unbalanced historical data. Data balancing processing is performed on the pre-acquired unbalanced historical data, which can effectively improve the distribution uniformity of the training data used for training the credit overdue risk prediction model, and can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, thereby effectively improving the accuracy and effectiveness of the credit overdue risk prediction, and providing banks and other financial institutions with a more accurate and reliable basis for judgment, so as to carry out early warning processing or risk control processing of credit overdue risks, thereby effectively reducing users' credit overdue risks and improving the user experience of financial institutions.

[0052] Based on the above content, the present application also provides a credit overdue risk prediction system for implementing the credit overdue risk prediction method provided in one or more embodiments of the present application. The credit overdue risk prediction system at least includes the credit overdue risk prediction device, the target processing system of the financial institution and the client device. First, in the model application stage, the credit overdue risk prediction device can receive the credit data of the target user from the client device held by the financial user, and then generate a corresponding target credit feature vector based on the credit data of the target user; input the target credit feature vector into a preset credit overdue risk prediction model, and determine the credit overdue risk prediction result of the target user based on the output of the credit overdue risk prediction model, and then send the credit overdue risk prediction result to the target processing system of the financial institution for early warning and display, and can also send the credit overdue risk prediction result to the client device of the risk control personnel of the financial institution. During the model training stage, the credit overdue risk prediction device can obtain historical credit data and labels corresponding to multiple historical users of the financial institution from the target processing system of the financial institution, and then, the labels include: a first label for indicating that the corresponding historical user has a credit overdue risk and a second label for indicating that the corresponding historical user does not have a credit overdue risk; generate a corresponding first data set according to the historical credit data and labels corresponding to each of the historical users, and divide the first data set into a training set and a test set; if the data in the training set is unbalanced historical data, perform data balancing processing on the training set to convert the data in the training set into balanced historical credit data and form a second data set; use the second data set to obtain the corresponding credit feature vector; and obtain the credit overdue risk prediction model based on the credit feature vector training.

[0053] In an actual application scenario, the credit overdue risk prediction device can be implemented using a server; the server can be communicatively connected to at least one client device.

[0054] It is understandable that the client device may include a smart phone, a tablet electronic device, a network set-top box, a portable computer, a desktop computer, a personal digital assistant (PDA), a vehicle-mounted device, a smart wearable device, etc. Among them, the smart wearable device may include smart glasses, a smart watch, a smart bracelet, etc.

[0055] In another practical application scenario, the part of the credit overdue risk prediction device that performs credit overdue risk prediction can be executed in the server as described above, or all operations can be completed in the client device. The specific selection can be made based on the processing capacity of the client device and the limitations of the user's usage scenario. This application does not limit this. If all operations are completed in the client device, the client device can also include a processor for specific processing of credit overdue risk prediction.

[0056] The client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side, and other implementation scenarios may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, or a server cluster consisting of multiple servers, or a server structure of a distributed device.

[0057] The server and the client device may communicate with each other using any suitable network protocol, including network protocols that have not yet been developed on the date of filing this application. The network protocols may include, for example, TCP / IP, UDP / IP, HTTP, HTTPS, etc. Of course, the network protocols may also include, for example, RPC (Remote Procedure Call Protocol) and REST (Representational State Transfer) protocols used on top of the above protocols.

[0058] The details are described in detail through the following embodiments and application examples.

[0059] In order to solve the problem that the accuracy of model training results is low due to unbalanced data, resulting in inaccurate credit overdue risk prediction results, the present application provides an embodiment of a credit overdue risk prediction method, see Figure 1 The credit overdue risk prediction method specifically includes the following contents:

[0060] Step 100: Generate a corresponding target credit feature vector based on the credit data of the target user.

[0061] In one or more embodiments of the present application, the credit data may specifically include customer credit data and basic information of financial users of financial institutions, and loan history detail data authorized by financial users, etc., which may be specifically set according to actual application scenarios.

[0062] In one or more embodiments of the present application, the target credit feature vector refers to a credit feature vector for a target user, wherein the credit feature vector refers to a feature vector that can be constructed by using expert experience to find important features and has a strong correlation with credit overdue prediction.

[0063] Step 200: Input the target credit feature vector into a preset credit overdue risk prediction model, and determine the credit overdue risk prediction result of the target user based on the output of the credit overdue risk prediction model; wherein the credit overdue risk prediction model is pre-trained based on balanced historical credit data, and the balanced historical credit data is data obtained after data balancing processing is performed on pre-acquired unbalanced historical data.

[0064] In one or more embodiments of the present application, the credit overdue risk prediction model can adopt a machine learning model capable of data prediction, preferably a random forest model. In machine learning, a random forest is a classifier containing multiple decision trees, and the category of its output is determined by the mode of the categories output by individual trees.

[0065] In step 200, balanced historical credit data and unbalanced historical data are relative. Specifically, the ratio or difference between the majority class data and the minority class data in the historical credit data can be used to determine whether the historical credit data belongs to unbalanced historical data or balanced historical credit data. For example, if the data without credit overdue risk in the current historical credit data accounts for 79%, and the data with credit overdue risk accounts for 21%, the ratio of the two is 79 / 21. Since the ratio is greater than the judgment threshold of 1.5, it means that the current historical credit data is unbalanced historical credit data, so it needs to be processed by data balancing to be transformed into balanced historical credit data. For example, after the unbalanced historical credit data is processed by data balancing, the data without credit overdue risk in the current historical credit data accounts for 55%, and the data with credit overdue risk accounts for 45%. The ratio of the two is 55 / 45. This ratio is less than the judgment threshold of 1.5, which means that the current historical credit data is balanced historical credit data.

[0066] It is understandable that the data balancing process may apply an existing data processing method that can interpolate the minority class and cluster or delete the majority class data.

[0067] From the above description, it can be seen that the credit overdue risk prediction method provided in the embodiment of the present application performs data balancing processing on the pre-acquired unbalanced historical data, which can effectively improve the distribution uniformity of the training data used for training the credit overdue risk prediction model, and can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, thereby effectively improving the accuracy and effectiveness of the credit overdue risk prediction, and providing banks and other financial institutions with a more accurate and reliable basis for judgment, so as to carry out early warning or risk control processing of credit overdue risks, thereby effectively reducing users' credit overdue risks and improving the user experience of financial institutions.

[0068] In order to provide a specific method for performing data balancing processing on the training set, in one embodiment of the credit overdue risk prediction method provided in the present application, see Figure 2 The credit overdue risk prediction method further specifically includes the following contents before step 200:

[0069] Step 010: Obtain historical credit data and labels corresponding to multiple historical users, wherein the labels include: a first label used to indicate that the corresponding historical user has a credit overdue risk and a second label used to indicate that the corresponding historical user does not have a credit overdue risk.

[0070] Specifically, customer samples required for modeling and analysis can be selected to obtain customer credit data, basic information, and loan history details authorized by the customer, and the credit data, basic information, and loan history details can be analyzed and converted into a data set.

[0071] Step 020: Generate a corresponding first data set according to the historical credit data and labels corresponding to each of the historical users, and divide the first data set into a training set and a test set.

[0072] In step 020, the first data set is specifically used to store the corresponding relationship between each of the historical users, historical credit data and tags.

[0073] Step 030: If the data in the training set is unbalanced historical data, data balancing is performed on the training set to convert the data in the training set into balanced historical credit data and form a second data set.

[0074] Step 040: Apply the second data set to obtain a corresponding credit feature vector.

[0075] Step 050: Obtain the credit overdue risk prediction model based on the credit feature vector training.

[0076] From the above description, it can be seen that the credit overdue risk prediction method provided in the embodiment of the present application can effectively improve the distribution uniformity of the training data used for training the credit overdue risk prediction model by performing data balancing processing on the training set, and can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model.

[0077] In order to provide a specific method of data preprocessing, in one embodiment of the credit overdue risk prediction method provided in this application, see Figure 3 The credit overdue risk prediction method further specifically includes the following contents between step 010 and step 020:

[0078] Step 011: pre-process the historical credit data corresponding to each of the historical users, wherein the pre-processing includes at least one of: completing missing data and deleting abnormal and invalid data.

[0079] From the above description, it can be seen that the credit overdue risk prediction method provided in the embodiment of the present application can effectively improve the application reliability and accuracy of historical credit data, so as to effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model.

[0080] In order to provide a specific method for judging unbalanced data, in one embodiment of the credit overdue risk prediction method provided in the present application, see Figure 4 The credit overdue risk prediction method further specifically includes the following contents between step 020 and step 030:

[0081] Step 021: If the ratio of the total number of historical users corresponding to the first label in the training set to the total number of historical users corresponding to the second label is greater than a first threshold or less than a second threshold, it is determined that the data in the training set is unbalanced historical data, wherein the first threshold is greater than 1 and the second threshold is greater than 0 and less than 1.

[0082] Specifically, if the majority class data is compared with the minority class data, the ratio needs to be compared with a first threshold, which is a value greater than 1, for example, the first threshold can be the aforementioned judgment threshold 1.5; if the minority class data is compared with the majority class data, the ratio needs to be compared with a second threshold, which is a value greater than 0 and less than 1, for example, the second threshold can be 0.75.

[0083] From the above description, it can be seen that the credit overdue risk prediction method provided in the embodiment of the present application can effectively improve the judgment efficiency and accuracy of unbalanced data, can effectively improve the efficiency and reliability of data balancing processing, and thus can effectively improve the efficiency and reliability of credit overdue risk prediction.

[0084] In order to provide a specific method of data balancing, in one embodiment of the credit overdue risk prediction method provided in this application, see Figure 5 Step 030 in the credit overdue risk prediction method specifically includes the following contents:

[0085] Step 031: Determine minority class data and majority class data in the training set according to the total number of historical users corresponding to the first label and the total number of historical users corresponding to the second label in the training set.

[0086] Step 032: Apply a preset data balancing method to perform K-nearest neighbor interpolation processing on the minority class data, and perform K-means clustering processing on the majority class data, so as to convert the data in the training set into balanced historical credit data and form a corresponding second data set; wherein the data balancing method includes: a preset SMOTE algorithm.

[0087] It is understandable that the SMOTE (Synthetic Minority Oversampling Technique) algorithm can preferably adopt a fusion SMOTE algorithm, which can be written as a KF-SMOTE algorithm, and can optimize a highly complex multi-dimensional unbalanced credit history data set. Among them, the synthetic minority oversampling technique is an improved solution based on the random oversampling algorithm. Since random oversampling adopts a simple sample duplication strategy to increase minority samples, it is easy to cause the problem of model overfitting, that is, the information learned by the model is too specific and not generalized enough. The basic idea of ​​the SMOTE algorithm is to analyze minority samples and artificially synthesize new samples based on minority samples and add them to the data set. The algorithm flow is as follows:

[0088] (1) For each sample x in the minority class, calculate its distance to all samples in the minority class sample set using the Euclidean distance as the criterion to obtain its k nearest neighbors.

[0089] (2) A sampling ratio is set according to the sample imbalance ratio to determine the sampling rate N. For each minority class sample x, several samples are randomly selected from its k nearest neighbors. Assume that the selected nearest neighbor is o.

[0090] (3) For each randomly selected neighbor o, a new sample is constructed with the original sample according to the following formula:

[0091] o(new)=o+rand(0,1)*(xo)

[0092] From the above description, it can be seen that the credit overdue risk prediction method provided in the embodiment of the present application can effectively improve the reliability and efficiency of data balancing processing, can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, and thus can effectively improve the accuracy and effectiveness of credit overdue risk prediction.

[0093] In order to provide a method for extracting feature vectors, in one embodiment of the credit overdue risk prediction method provided in the present application, step 040 in the credit overdue risk prediction method specifically includes the following contents:

[0094] Step 041: Perform data dimension reduction and feature screening processing on the second data set to obtain the credit feature vectors corresponding to each of the historical users and the virtual users obtained through data balancing processing.

[0095] From the above description, it can be seen that the credit overdue risk prediction method provided in the embodiment of the present application can effectively improve the reliability and efficiency of feature vector extraction, can effectively improve the reliability and efficiency of the trained credit overdue risk prediction model, and thus can effectively improve the reliability and efficiency of credit overdue risk prediction.

[0096] In order to provide a specific method for model training, in one embodiment of the credit overdue risk prediction method provided in this application, see Figure 6 Step 050 in the credit overdue risk prediction method specifically includes the following contents:

[0097] Step 051: Apply a preset random forest algorithm to obtain an initial prediction model based on the credit feature vector training.

[0098] Step 052: Evaluate the effectiveness of the initial prediction model according to the evaluation index corresponding to the test set, and adjust the initial prediction model based on the corresponding effect evaluation result to obtain a corresponding credit overdue risk prediction model.

[0099] From the above description, it can be seen that the credit overdue risk prediction method provided in the embodiment of the present application can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, and thus can effectively improve the accuracy and effectiveness of the credit overdue risk prediction.

[0100] From the software level, in order to solve the problem that the accuracy of the model training results is low due to unbalanced data, resulting in inaccurate credit overdue risk prediction results, the present application provides an embodiment of a credit overdue risk prediction device for executing all or part of the content of the credit overdue risk prediction method, see Figure 7 The credit overdue risk prediction device specifically includes the following contents:

[0101] The vector generation module 10 is used to generate a corresponding target credit feature vector based on the credit data of the target user.

[0102] In one or more embodiments of the present application, the credit data may specifically include customer credit data and basic information of financial users of financial institutions, and loan history detail data authorized by financial users, etc., which may be specifically set according to actual application scenarios.

[0103] In one or more embodiments of the present application, the target credit feature vector refers to a credit feature vector for a target user, wherein the credit feature vector refers to a feature vector that can be constructed by using expert experience to find important features and has a strong correlation with credit overdue prediction.

[0104] The model prediction module 20 is used to input the target credit feature vector into a preset credit overdue risk prediction model, and determine the credit overdue risk prediction result of the target user based on the output of the credit overdue risk prediction model; wherein the credit overdue risk prediction model is pre-trained based on balanced historical credit data, and the balanced historical credit data is data obtained after data balancing processing is performed on pre-acquired unbalanced historical data.

[0105] In one or more embodiments of the present application, the credit overdue risk prediction model can adopt a machine learning model capable of data prediction, preferably a random forest model. In machine learning, a random forest is a classifier containing multiple decision trees, and the category of its output is determined by the mode of the categories output by individual trees.

[0106] In the model prediction module 20, balanced historical credit data and unbalanced historical data are relative. Specifically, the ratio or difference between the majority class data and the minority class data in the historical credit data can be used to determine whether the historical credit data belongs to unbalanced historical data or balanced historical credit data. For example: if the data without credit overdue risk in the current historical credit data accounts for 79%, and the data with credit overdue risk accounts for 21%, the ratio of the two is 79 / 21. Since the ratio is greater than the judgment threshold of 1.5, it means that the current historical credit data is unbalanced historical credit data, so it needs to be processed by data balancing to be transformed into balanced historical credit data. For example, after the unbalanced historical credit data is processed by data balancing, the data without credit overdue risk in the current historical credit data accounts for 55%, and the data with credit overdue risk accounts for 45%. The ratio of the two is 55 / 45. This ratio is less than the judgment threshold of 1.5, which means that the current historical credit data is balanced historical credit data.

[0107] It is understandable that the data balancing process may apply an existing data processing method that can interpolate the minority class and cluster or delete the majority class data.

[0108] The embodiment of the credit overdue risk prediction device provided in the present application can be specifically used to execute the processing flow of the embodiment of the credit overdue risk prediction method in the above embodiment. Its functions will not be repeated here, and reference can be made to the detailed description of the above method embodiment.

[0109] From the above description, it can be seen that the credit overdue risk prediction device provided in the embodiment of the present application performs data balancing processing on the pre-acquired unbalanced historical data, which can effectively improve the distribution uniformity of the training data used for training the credit overdue risk prediction model, and can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, thereby effectively improving the accuracy and effectiveness of the credit overdue risk prediction, and providing banks and other financial institutions with a more accurate and reliable basis for judgment, so as to carry out early warning or risk control processing of credit overdue risks, thereby effectively reducing users' credit overdue risks and improving the user experience of financial institutions.

[0110] In order to provide a specific method for performing data balancing processing on the training set, in one embodiment of the credit overdue risk prediction device provided in the present application, see Figure 8 The credit overdue risk prediction device further specifically includes a model training module 01, which is used to execute the following contents:

[0111] Step 010: Obtain historical credit data and labels corresponding to multiple historical users, wherein the labels include: a first label used to indicate that the corresponding historical user has a credit overdue risk and a second label used to indicate that the corresponding historical user does not have a credit overdue risk.

[0112] Specifically, customer samples required for modeling and analysis can be selected to obtain customer credit data, basic information, and loan history details authorized by the customer, and the credit data, basic information, and loan history details can be analyzed and converted into a data set.

[0113] Step 020: Generate a corresponding first data set according to the historical credit data and labels corresponding to each of the historical users, and divide the first data set into a training set and a test set.

[0114] In step 020, the first data set is specifically used to store the corresponding relationship between each of the historical users, historical credit data and tags.

[0115] Step 030: If the data in the training set is unbalanced historical data, data balancing is performed on the training set to convert the data in the training set into balanced historical credit data and form a second data set.

[0116] Step 040: Apply the second data set to obtain a corresponding credit feature vector.

[0117] Step 050: Obtain the credit overdue risk prediction model based on the credit feature vector training.

[0118] From the above description, it can be seen that the credit overdue risk prediction device provided in the embodiment of the present application can effectively improve the distribution uniformity of the training data used for training the credit overdue risk prediction model by performing data balancing processing on the training set, and can effectively improve the accuracy, reliability and effectiveness of the credit overdue risk prediction model obtained by training.

[0119] In order to provide a specific method of data preprocessing, in one embodiment of the credit overdue risk prediction device provided in the present application, the model training module 01 in the credit overdue risk prediction device is also used to perform the following contents:

[0120] Step 011: pre-process the historical credit data corresponding to each of the historical users, wherein the pre-processing includes at least one of: completing missing data and deleting abnormal and invalid data.

[0121] From the above description, it can be seen that the credit overdue risk prediction device provided in the embodiment of the present application can effectively improve the application reliability and accuracy of historical credit data, so as to effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model.

[0122] In order to provide a specific method for judging unbalanced data, in one embodiment of the credit overdue risk prediction device provided in the present application, the model training module 01 in the credit overdue risk prediction device is also used to perform the following contents:

[0123] Step 021: If the ratio of the total number of historical users corresponding to the first label in the training set to the total number of historical users corresponding to the second label is greater than a first threshold or less than a second threshold, it is determined that the data in the training set is unbalanced historical data, wherein the first threshold is greater than 1 and the second threshold is greater than 0 and less than 1.

[0124] Specifically, if the majority class data is compared with the minority class data, the ratio needs to be compared with a first threshold, which is a value greater than 1, for example, the first threshold can be the aforementioned judgment threshold 1.5; if the minority class data is compared with the majority class data, the ratio needs to be compared with a second threshold, which is a value greater than 0 and less than 1, for example, the second threshold can be 0.75.

[0125] From the above description, it can be seen that the credit overdue risk prediction device provided in the embodiment of the present application can effectively improve the judgment efficiency and accuracy of unbalanced data, can effectively improve the efficiency and reliability of data balancing processing, and thus can effectively improve the efficiency and reliability of credit overdue risk prediction.

[0126] In order to provide a specific method of data balancing, in one embodiment of the credit overdue risk prediction device provided in the present application, the model training module 01 in the credit overdue risk prediction device is also used to perform the following contents:

[0127] Step 031: Determine minority class data and majority class data in the training set according to the total number of historical users corresponding to the first label and the total number of historical users corresponding to the second label in the training set.

[0128] Step 032: Apply a preset data balancing method to perform K-nearest neighbor interpolation processing on the minority class data, and perform K-means clustering processing on the majority class data, so as to convert the data in the training set into balanced historical credit data and form a corresponding second data set; wherein the data balancing method includes: a preset SMOTE algorithm.

[0129] It is understandable that the SMOTE (Synthetic Minority Oversampling Technique) algorithm can preferably adopt a fusion SMOTE algorithm, which can be written as a KF-SMOTE algorithm, and can optimize a highly complex multi-dimensional unbalanced credit history data set. Among them, the synthetic minority oversampling technique is an improved solution based on the random oversampling algorithm. Since random oversampling adopts a simple sample duplication strategy to increase minority samples, it is easy to cause the problem of model overfitting, that is, the information learned by the model is too specific and not generalized enough. The basic idea of ​​the SMOTE algorithm is to analyze minority samples and artificially synthesize new samples based on minority samples and add them to the data set. The algorithm flow is as follows:

[0130] (1) For each sample x in the minority class, calculate its distance to all samples in the minority class sample set using the Euclidean distance as the criterion to obtain its k nearest neighbors.

[0131] (2) A sampling ratio is set according to the sample imbalance ratio to determine the sampling rate N. For each minority class sample x, several samples are randomly selected from its k nearest neighbors. Assume that the selected nearest neighbor is o.

[0132] (3) For each randomly selected neighbor o, a new sample is constructed with the original sample according to the following formula:

[0133] o(new)=o+rand(0,1)*(xo)

[0134] From the above description, it can be seen that the credit overdue risk prediction device provided in the embodiment of the present application can effectively improve the reliability and efficiency of data balancing processing, can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, and thus can effectively improve the accuracy and effectiveness of credit overdue risk prediction.

[0135] In order to provide a method for extracting feature vectors, in one embodiment of the credit overdue risk prediction device provided in the present application, the model training module 01 in the credit overdue risk prediction device is also used to perform the following:

[0136] Step 041: Perform data dimension reduction and feature screening processing on the second data set to obtain the credit feature vectors corresponding to each of the historical users and the virtual users obtained through data balancing processing.

[0137] From the above description, it can be seen that the credit overdue risk prediction device provided in the embodiment of the present application can effectively improve the reliability and efficiency of feature vector extraction, can effectively improve the reliability and efficiency of the trained credit overdue risk prediction model, and thus can effectively improve the reliability and efficiency of credit overdue risk prediction.

[0138] In order to provide a specific method of model training, in one embodiment of the credit overdue risk prediction device provided in the present application, the model training module 01 in the credit overdue risk prediction device is also used to perform the following contents:

[0139] Step 051: Apply a preset random forest algorithm to obtain an initial prediction model based on the credit feature vector training.

[0140] Step 052: Evaluate the effectiveness of the initial prediction model according to the evaluation index corresponding to the test set, and adjust the initial prediction model based on the corresponding effect evaluation result to obtain a corresponding credit overdue risk prediction model.

[0141] From the above description, it can be seen that the credit overdue risk prediction device provided in the embodiment of the present application can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, and thus can effectively improve the accuracy and effectiveness of the credit overdue risk prediction.

[0142] In order to further illustrate this scheme, this application also provides a specific application example of a credit overdue risk prediction method implemented by a credit overdue risk prediction device. In this application example, it includes the acquisition of sample data sets, preprocessing of unbalanced data sets, feature engineering, training and parameter adjustment of risk prediction models, and evaluation of model performance, which realizes intelligent credit risk assessment and prediction of credit application customers, and mainly improves the shortcomings of low accuracy of prediction models due to uneven data, thereby improving the performance of prediction models. The main technical innovation is that a fusion SMOTE algorithm of minority class K nearest neighbor interpolation and majority class K mean clustering is proposed, namely KF-SMOTE, which can optimize highly complex multi-dimensional unbalanced credit historical data sets, ensure the performance of credit risk assessment prediction models, and provide a more scientific reference for bank credit system construction and risk management.

[0143] See also Fig. 9 The application example of this application provides a credit overdue risk prediction method for unbalanced data sets, including: sample data acquisition, data preprocessing, proposing a KF-SMOTE algorithm to solve the problem of unbalanced data sets, feature engineering, a credit overdue risk assessment model based on a random forest algorithm, and model evaluation. Specifically, it includes the following steps:

[0144] Step 1): Sample data acquisition: select the customer samples required for modeling and analysis, obtain customer credit data, basic information and loan history details authorized by the customer, analyze and convert the credit data, basic information and loan history details into a data set.

[0145] Step 2): Data preprocessing, summarize and organize the data obtained in step 1), including the processing of missing, abnormal and invalid values.

[0146] Step 3): See Fig.10 , the processing of unbalanced data sets, first divides the training set and the test set, and then applies the proposed KF-SMOTE algorithm to the training set, performs K nearest neighbor interpolation on the minority class data in the preprocessed data set, and performs K means clustering on the majority class data, thereby completing the homogenization of the unbalanced data set and obtaining a training set with uniform data; wherein the KF-SMOTE algorithm performs clustering and interpolation processing on the majority class data and the minority class data respectively.

[0147] Step 4): Feature engineering: perform data dimension reduction and feature screening on the preprocessed data, use expert experience to find important features, and construct feature vectors that are highly correlated with credit delinquency prediction.

[0148] Step 5): Build a risk prediction model and use the random forest algorithm to train on the training data set; repeatedly adjust the adjustable parameters of the model until the model with the best performance is obtained; finally, use the test set to obtain the prediction model evaluation index.

[0149] Step 6): Model evaluation: Through model evaluation indicators and comparison with other algorithm models, the model's discrimination ability, prediction ability and stability are comprehensively evaluated.

[0150] From the above description, it can be seen that in order to solve the model failure problem caused by data imbalance in the existing credit overdue risk prediction method, the credit overdue risk prediction method provided in the application example of this application preprocesses the unbalanced modeling sample data through the proposed KF-SMOTE algorithm, thereby improving the accuracy of the prediction model, thereby providing a guarantee for establishing a reasonable and effective credit overdue risk prediction model. Its advantages are as follows:

[0151] 1. The proposed KF-SMOTE algorithm optimizes the highly complex multi-dimensional unbalanced credit data set and ensures the reliability of the credit risk assessment prediction model;

[0152] 2. The establishment of the risk prediction model is combined with the integrated random forest algorithm, which makes the model have strong generalization ability and training speed.

[0153] From the hardware level, in order to solve the problem of low accuracy of model training results caused by unbalanced data and thus inaccurate credit overdue risk prediction results, the present application provides an embodiment of an electronic device for implementing all or part of the content of the credit overdue risk prediction method, and the electronic device specifically includes the following content:

[0154] Fig.11 FIG. 9 is a schematic block diagram of the system structure of the electronic device 9600 according to an embodiment of the present application. Fig.11 As shown, the electronic device 9600 may include a central processor 9100 and a memory 9140; the memory 9140 is coupled to the central processor 9100. It is worth noting that Fig.11 is exemplary; other types of structures may also be used to supplement or replace this structure to implement telecommunication functions or other functions.

[0155] In one embodiment, the credit overdue risk prediction function may be integrated into a central processing unit, wherein the central processing unit may be configured to perform the following control:

[0156] Step 100: Generate a corresponding target credit feature vector based on the credit data of the target user.

[0157] In one or more embodiments of the present application, the credit data may specifically include customer credit data and basic information of financial users of financial institutions, and loan history detail data authorized by financial users, etc., which may be specifically set according to actual application scenarios.

[0158] In one or more embodiments of the present application, the target credit feature vector refers to a credit feature vector for a target user, wherein the credit feature vector refers to a feature vector that can be constructed by using expert experience to find important features and has a strong correlation with credit overdue prediction.

[0159] Step 200: Input the target credit feature vector into a preset credit overdue risk prediction model, and determine the credit overdue risk prediction result of the target user based on the output of the credit overdue risk prediction model; wherein the credit overdue risk prediction model is pre-trained based on balanced historical credit data, and the balanced historical credit data is data obtained after data balancing processing is performed on pre-acquired unbalanced historical data.

[0160] In one or more embodiments of the present application, the credit overdue risk prediction model can adopt a machine learning model capable of data prediction, preferably a random forest model. In machine learning, a random forest is a classifier containing multiple decision trees, and the category of its output is determined by the mode of the categories output by individual trees.

[0161] In step 200, balanced historical credit data and unbalanced historical data are relative. Specifically, the ratio or difference between the majority class data and the minority class data in the historical credit data can be used to determine whether the historical credit data belongs to unbalanced historical data or balanced historical credit data. For example, if the data without credit overdue risk in the current historical credit data accounts for 79%, and the data with credit overdue risk accounts for 21%, the ratio of the two is 79 / 21. Since the ratio is greater than the judgment threshold of 1.5, it means that the current historical credit data is unbalanced historical credit data, so it needs to be processed by data balancing to be transformed into balanced historical credit data. For example, after the unbalanced historical credit data is processed by data balancing, the data without credit overdue risk in the current historical credit data accounts for 55%, and the data with credit overdue risk accounts for 45%. The ratio of the two is 55 / 45. This ratio is less than the judgment threshold of 1.5, which means that the current historical credit data is balanced historical credit data.

[0162] It is understandable that the data balancing process may apply an existing data processing method that can interpolate the minority class and cluster or delete the majority class data.

[0163] From the above description, it can be seen that the electronic device provided in the embodiment of the present application performs data balancing processing on the pre-acquired unbalanced historical data, which can effectively improve the distribution uniformity of the training data used for training the credit overdue risk prediction model, and can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, thereby effectively improving the accuracy and effectiveness of the credit overdue risk prediction, and providing banks and other financial institutions with a more accurate and reliable basis for judgment, so as to carry out early warning or risk control processing of credit overdue risks, thereby effectively reducing the user's credit overdue risk and improving the user experience of financial institutions.

[0164] In another embodiment, the credit overdue risk prediction device can be configured separately from the central processor 9100. For example, the credit overdue risk prediction device can be configured as a chip connected to the central processor 9100, and the credit overdue risk prediction function can be realized through the control of the central processor.

[0165] like Fig.11 As shown, the electronic device 9600 may also include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily have to include Fig.11 In addition, the electronic device 9600 may also include Fig.11 For components not shown, reference may be made to the prior art.

[0166] like Fig.11 As shown, the central processing unit 9100 is sometimes also referred to as a controller or an operation control, and may include a microprocessor or other processor device and / or logic device. The central processing unit 9100 receives input and controls the operation of various components of the electronic device 9600.

[0167] The memory 9140 may be, for example, one or more of a cache, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory or other suitable devices. The above-mentioned information related to the failure may be stored, and a program for executing the relevant information may also be stored. The CPU 9100 may execute the program stored in the memory 9140 to implement information storage or processing, etc.

[0168] The input unit 9120 provides input to the central processing unit 9100. The input unit 9120 is, for example, a key or a touch input device. The power supply 9170 is used to provide power to the electronic device 9600. The display 9160 is used to display display objects such as images and texts. The display may be, for example, an LCD display, but is not limited thereto.

[0169] The memory 9140 may be a solid-state memory, such as a read-only memory (ROM), a random access memory (RAM), a SIM card, etc. It may also be a memory that saves information even when the power is off, can be selectively erased, and is provided with more data, examples of which are sometimes referred to as EPROMs, etc. The memory 9140 may also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142, which is used to store application programs and function programs or processes for executing the operation of the electronic device 9600 through the central processor 9100.

[0170] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for communication functions of the electronic device and / or for executing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0171] The communication module 9110 is a transmitter / receiver 9110 that sends and receives signals via an antenna 9111. The communication module (transmitter / receiver) 9110 is coupled to the central processor 9100 to provide input signals and receive output signals, which may be the same as the case of a conventional mobile communication terminal.

[0172] Based on different communication technologies, multiple communication modules 9110 may be provided in the same electronic device, such as a cellular network module, a Bluetooth module and / or a wireless LAN module, etc. The communication module (transmitter / receiver) 9110 is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide an audio output via the speaker 9131 and receive an audio input from the microphone 9132, thereby realizing a common telecommunication function. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. In addition, the audio processor 9130 is also coupled to the central processor 9100, so that recording can be performed on the local machine through the microphone 9132, and the sound stored on the local machine can be played through the speaker 9131.

[0173] The embodiments of the present application also provide a computer-readable storage medium capable of implementing all the steps of the credit overdue risk prediction method in the above embodiments. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, all the steps of the credit overdue risk prediction method in the above embodiments are implemented. For example, when the processor executes the computer program, the following steps are implemented:

[0174] Step 100: Generate a corresponding target credit feature vector based on the credit data of the target user.

[0175] In one or more embodiments of the present application, the credit data may specifically include customer credit data and basic information of financial users of financial institutions, and loan history detail data authorized by financial users, etc., which may be specifically set according to actual application scenarios.

[0176] In one or more embodiments of the present application, the target credit feature vector refers to a credit feature vector for a target user, wherein the credit feature vector refers to a feature vector that can be constructed by using expert experience to find important features and has a strong correlation with credit overdue prediction.

[0177] Step 200: Input the target credit feature vector into a preset credit overdue risk prediction model, and determine the credit overdue risk prediction result of the target user based on the output of the credit overdue risk prediction model; wherein the credit overdue risk prediction model is pre-trained based on balanced historical credit data, and the balanced historical credit data is data obtained after data balancing processing is performed on pre-acquired unbalanced historical data.

[0178] In one or more embodiments of the present application, the credit overdue risk prediction model can adopt a machine learning model capable of data prediction, preferably a random forest model. In machine learning, a random forest is a classifier containing multiple decision trees, and the category of its output is determined by the mode of the categories output by individual trees.

[0179] In step 200, balanced historical credit data and unbalanced historical data are relative. Specifically, the ratio or difference between the majority class data and the minority class data in the historical credit data can be used to determine whether the historical credit data belongs to unbalanced historical data or balanced historical credit data. For example, if the data without credit overdue risk in the current historical credit data accounts for 79%, and the data with credit overdue risk accounts for 21%, the ratio of the two is 79 / 21. Since the ratio is greater than the judgment threshold of 1.5, it means that the current historical credit data is unbalanced historical credit data, so it needs to be processed by data balancing to be transformed into balanced historical credit data. For example, after the unbalanced historical credit data is processed by data balancing, the data without credit overdue risk in the current historical credit data accounts for 55%, and the data with credit overdue risk accounts for 45%. The ratio of the two is 55 / 45. This ratio is less than the judgment threshold of 1.5, which means that the current historical credit data is balanced historical credit data.

[0180] It is understandable that the data balancing process may apply an existing data processing method that can interpolate the minority class and cluster or delete the majority class data.

[0181] From the above description, it can be seen that the computer-readable storage medium provided in the embodiment of the present application performs data balancing processing on the pre-acquired unbalanced historical data, which can effectively improve the distribution uniformity of the training data used for training the credit overdue risk prediction model, and can effectively improve the accuracy, reliability and effectiveness of the trained credit overdue risk prediction model, thereby effectively improving the accuracy and effectiveness of the credit overdue risk prediction, and providing banks and other financial institutions with a more accurate and reliable basis for judgment, so as to carry out early warning or risk control processing of credit overdue risks, thereby effectively reducing users' credit overdue risks and improving the user experience of financial institutions.

[0182] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (devices), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the functions in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0183] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0184] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A method for predicting credit overdue risk, characterized in that: include: Generate a corresponding target credit feature vector based on the target user's credit data, wherein the credit data includes: customer credit data and basic information of the financial user of the financial institution and loan history detailed data authorized by the financial user; Inputting the target credit feature vector into a preset credit overdue risk prediction model, and determining a credit overdue risk prediction result of the target user based on an output of the credit overdue risk prediction model; The credit overdue risk prediction model is pre-trained based on balanced historical credit data, and the balanced historical credit data is data obtained after data balancing of pre-acquired unbalanced historical data, and the unbalanced historical data and balanced historical credit data are distinguished according to the ratio or difference between the majority class data and the minority class data in the historical credit data; Before inputting the target credit feature vector into a preset credit overdue risk prediction model, the method further includes: Acquire historical credit data and labels corresponding to each of a plurality of historical users, wherein the labels include: a first label for indicating that the corresponding historical user has a credit overdue risk and a second label for indicating that the corresponding historical user does not have a credit overdue risk; Generate a corresponding first data set according to the historical credit data and labels corresponding to each of the historical users, and divide the first data set into a training set and a test set; If the data in the training set is unbalanced historical data, performing data balancing processing on the training set so as to convert the data in the training set into balanced historical credit data and form a second data set; Applying the second data set to obtain a corresponding credit feature vector; The credit overdue risk prediction model is obtained based on the credit feature vector training; Before performing data balancing processing on the training set to convert the data in the training set into balanced historical credit data and form a second data set if the data in the training set is unbalanced historical data, the method further includes: If the ratio of the total number of historical users corresponding to the first label in the training set to the total number of historical users corresponding to the second label is greater than a first threshold or less than a second threshold, it is determined that the data in the training set is unbalanced historical data, wherein the first threshold is greater than 1 and the second threshold is greater than 0 and less than 1; The performing data balancing processing on the training set so as to convert the data in the training set into balanced historical credit data and form a second data set includes: Determining minority class data and majority class data in the training set according to the total number of historical users corresponding to the first label and the total number of historical users corresponding to the second label in the training set; Applying a preset data balancing method to perform K nearest neighbor interpolation processing on the minority class data, and performing K mean clustering processing on the majority class data, so as to convert the data in the training set into balanced historical credit data and form a corresponding second data set; Wherein, the data balancing method includes: a preset SMOTE algorithm; If the ratio of the total number of historical users corresponding to the first label in the training set to the total number of historical users corresponding to the second label is greater than a first threshold or less than a second threshold, determining that the data in the training set is unbalanced historical data includes: If the majority class data is compared with the minority class data, the ratio is compared with the first threshold. If it is greater than the first threshold, it is determined that the data in the training set is unbalanced historical data, and the first threshold is 1.5; if the minority class data is compared with the majority class data, the ratio is compared with the second threshold. If it is less than the second threshold, it is determined that the data in the training set is unbalanced historical data, and the second threshold is 0.

75.

2. The credit overdue risk prediction method according to claim 1, characterized in that: Before dividing the first data set into a training set and a test set, the method further includes: The historical credit data corresponding to each of the historical users is preprocessed, wherein the preprocessing includes at least one of: completing missing data and deleting abnormal and invalid data.

3. The credit overdue risk prediction method according to claim 1, characterized in that: The applying the second data set to obtain a corresponding credit feature vector includes: The second data set is subjected to data dimension reduction and feature screening processing to obtain credit feature vectors corresponding to each of the historical users and the virtual users obtained through data balancing processing.

4. The credit overdue risk prediction method according to claim 1, characterized in that: The credit overdue risk prediction model obtained based on the credit feature vector training includes: Applying a preset random forest algorithm to obtain an initial prediction model based on the credit feature vector training; The initial prediction model is evaluated for effectiveness according to the evaluation index corresponding to the test set, and the initial prediction model is adjusted based on the corresponding effect evaluation result to obtain the corresponding credit overdue risk prediction model.

5. A credit overdue risk prediction device, characterized in that: include: A vector generation module, used to generate a corresponding target credit feature vector based on the credit data of the target user, wherein the credit data includes: customer credit data and basic information of the financial user of the financial institution and loan history detailed data authorized by the financial user; A model prediction module, used for inputting the target credit feature vector into a preset credit overdue risk prediction model, and determining the credit overdue risk prediction result of the target user based on the output of the credit overdue risk prediction model; The credit overdue risk prediction model is pre-trained based on balanced historical credit data, and the balanced historical credit data is data obtained after data balancing of pre-acquired unbalanced historical data, and the unbalanced historical data and balanced historical credit data are distinguished according to the ratio or difference between the majority class data and the minority class data in the historical credit data; Before inputting the target credit feature vector into a preset credit overdue risk prediction model, the method further includes: Acquire historical credit data and labels corresponding to each of a plurality of historical users, wherein the labels include: a first label for indicating that the corresponding historical user has a credit overdue risk and a second label for indicating that the corresponding historical user does not have a credit overdue risk; Generate a corresponding first data set according to the historical credit data and labels corresponding to each of the historical users, and divide the first data set into a training set and a test set; If the data in the training set is unbalanced historical data, performing data balancing processing on the training set so as to convert the data in the training set into balanced historical credit data and form a second data set; Applying the second data set to obtain a corresponding credit feature vector; The credit overdue risk prediction model is obtained based on the credit feature vector training; Before performing data balancing processing on the training set to convert the data in the training set into balanced historical credit data and form a second data set if the data in the training set is unbalanced historical data, the method further includes: If the ratio of the total number of historical users corresponding to the first label in the training set to the total number of historical users corresponding to the second label is greater than a first threshold or less than a second threshold, it is determined that the data in the training set is unbalanced historical data, wherein the first threshold is greater than 1 and the second threshold is greater than 0 and less than 1; The performing data balancing processing on the training set so as to convert the data in the training set into balanced historical credit data and form a second data set includes: Determining minority class data and majority class data in the training set according to the total number of historical users corresponding to the first label and the total number of historical users corresponding to the second label in the training set; Applying a preset data balancing method to perform K nearest neighbor interpolation processing on the minority class data, and performing K mean clustering processing on the majority class data, so as to convert the data in the training set into balanced historical credit data and form a corresponding second data set; Wherein, the data balancing method includes: a preset SMOTE algorithm; If the ratio of the total number of historical users corresponding to the first label in the training set to the total number of historical users corresponding to the second label is greater than a first threshold or less than a second threshold, determining that the data in the training set is unbalanced historical data includes: If the majority class data is compared with the minority class data, the ratio is compared with the first threshold. If it is greater than the first threshold, it is determined that the data in the training set is unbalanced historical data, and the first threshold is 1.5; if the minority class data is compared with the majority class data, the ratio is compared with the second threshold. If it is less than the second threshold, it is determined that the data in the training set is unbalanced historical data, and the second threshold is 0.

75.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the credit overdue risk prediction method according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the credit overdue risk prediction method described in any one of claims 1 to 4 is implemented.

Images