Credit risk detection method and device, equipment and storage medium
The high-dimensional data in credit risk analysis is reduced through principal element analysis technology, which solves the problem of high complexity in model training and calculation, and improves model training efficiency and detection accuracy.
Patent Information
- Application Number
- CN202411732849.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-28
- Publication Date
- 2025-02-18
Smart Images
Figure CN119477526A_ABST
Abstract
Description
Technical Field
[0001] Embodiments of the present invention relate to the technical field of credit risk analysis, and in particular to a credit risk detection method, device, equipment and storage medium. Background Art
[0002] In modern finance, credit risk management faces a series of practical problems, which include not only challenges from the financial market but also technical and methodological limitations.
[0003] First, the uncertainty and surge in competition in the financial market have made credit risk management more complex. Traditional risk assessment methods based on rules and experience often have difficulty in coping with the challenges brought about by market changes. Therefore, traditional anomaly identification methods such as decision trees and support vector machines have emerged. These methods use historical data and customer information to establish credit risk identification models to determine whether customers are likely to default. However, these methods also have certain shortcomings. With the explosive growth of transactions, the high dimensionality and complexity of data have brought new challenges to traditional risk anomaly identification methods. Financial institutions usually have a large amount of customer data and transaction data. How to extract effective information from these massive data and accurately assess risks is an urgent problem to be solved. Traditional anomaly identification methods cannot handle the collinearity problem in customer credit data and it is difficult to capture the potential patterns and laws in the data; when faced with massive high-dimensional data, they will face the problem of dimensional disaster, and the model training and computational complexity will increase, which greatly limits their application in credit risk management. Summary of the invention
[0004] The embodiments of the present invention provide a credit risk detection method, device, equipment and storage medium. The technical solution of the embodiments of the present invention solves the problems of complex sample data and high computational complexity of model training in existing credit risk analysis technologies. The high-dimensional sample data can be reduced in dimension through principal component analysis technology, and model training can be performed based on the reduced-dimensional sample data, thereby reducing the computational complexity of model training and improving the efficiency of model training.
[0005] In a first aspect, an embodiment of the present invention provides a credit risk detection method, the method comprising:
[0006] Obtain the target object's credit information to be analyzed; input the credit information to be analyzed into a pre-trained target risk detection model to obtain a target risk detection result of the target object; wherein the target risk detection model is trained based on preset sample information after principal component analysis.
[0007] In a second aspect, an embodiment of the present invention provides a credit risk detection device, the device comprising:
[0008] A credit information acquisition module is used to obtain the credit information to be analyzed of the target object; a credit information analysis module is used to input the credit information to be analyzed into a pre-trained target risk detection model to obtain the target risk detection result of the target object; wherein the target risk detection model is trained based on preset sample information after principal component analysis.
[0009] In a third aspect, an embodiment of the present invention provides a computer device, the computer device comprising:
[0010] one or more processors;
[0011] A memory for storing one or more programs;
[0012] When the one or more programs are executed by the one or more processors, the one or more processors implement the credit risk detection method described in any embodiment.
[0013] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the credit risk detection method described in any embodiment.
[0014] The technical solution provided by the embodiment of the present invention obtains the credit information to be analyzed of the target object; inputs the credit information to be analyzed into a pre-trained target risk detection model to obtain the target risk detection result of the target object; wherein the target risk detection model is trained based on preset sample information after principal component analysis. The technical solution of the embodiment of the present invention solves the problems of complex sample data and high computational complexity of model training in existing credit risk analysis technologies. The principal component analysis technology can be used to reduce the dimensionality of high-dimensional sample data, and model training can be performed based on the reduced-dimensional sample data, thereby reducing the computational complexity of model training and improving the efficiency of model training. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 is a flow chart of a credit risk detection method provided by an embodiment of the present invention;
[0016] Figure 2 is a flow chart of another credit risk detection method provided by an embodiment of the present invention;
[0017] Figure 3 It is a structural schematic diagram of a credit risk detection device provided by an embodiment of the present invention;
[0018] Figure 4 It is a structural schematic diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0019] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0020] Figure 1 It is a flow chart of a credit risk detection method provided by an embodiment of the present invention. The embodiment of the present invention can be applied to the scenario of detecting the credit risk of a target object. The method can be executed by a credit risk detection device, which can be implemented by software and / or hardware.
[0021] like Figure 1 As shown, the credit risk detection method includes the following steps:
[0022] S110, obtaining the credit information to be analyzed of the target object.
[0023] The target object may be an object that needs to be subjected to credit risk detection. The credit information to be analyzed may be credit information that needs to be subjected to credit risk analysis. Specifically, the historical credit information of the target object may be obtained based on a preset acquisition method, and the historical credit information of the target object may be used as the credit information to be analyzed.
[0024] S120: input the credit information to be analyzed into a pre-trained target risk detection model to obtain a target risk detection result of the target object.
[0025] The target risk detection model may be a model that is ultimately used to detect the credit risk of the target object. The target risk detection model is trained based on preset sample information after principal component analysis. The target risk detection result may be a credit risk detection result for the target object. Exemplarily, the target risk detection result may include detection results of types such as risk-free and risky. Specifically, the credit information to be analyzed may be input into a pre-trained target risk detection model so that the model analyzes the credit information to be analyzed, and then obtains the target risk detection result of the target object.
[0026] The technical solution provided by the embodiment of the present invention obtains the credit information to be analyzed of the target object; inputs the credit information to be analyzed into a pre-trained target risk detection model to obtain the target risk detection result of the target object; wherein the target risk detection model is trained based on preset sample information after principal component analysis. The technical solution of the embodiment of the present invention solves the problems of complex sample data and high computational complexity of model training in existing credit risk analysis technologies. The principal component analysis technology can be used to reduce the dimensionality of high-dimensional sample data, and model training can be performed based on the reduced-dimensional sample data, thereby reducing the computational complexity of model training and improving the efficiency of model training.
[0027] Figure 2 It is another flow chart of a credit risk detection method provided by an embodiment of the present invention. The embodiment of the present invention can be applied to the scenario of detecting the credit risk of a target object. Based on the above embodiment, this embodiment further illustrates how to train a target risk detection model; and how to detect the credit risk of a target object based on the target risk detection model. The device can be implemented by software and / or hardware and integrated into a computer device with application development function.
[0028] like Figure 2 As shown, the credit risk detection method includes the following steps:
[0029] S210: Obtain a preliminary risk detection model and preset sample information.
[0030] The preliminary risk detection model may be an original credit risk detection model that has not been pre-trained. Furthermore, the preset sample information may be preset sample information about the target object. Exemplarily, a certain amount of credit information data may be selected from a preset credit information database as the preset sample information. The preset sample information may contain corresponding risk labels for evaluating the detection results of the initial risk detection model and adjusting the model according to the evaluation results.
[0031] S220: Perform principal component analysis on the preset sample information to obtain processed sample information.
[0032] The processed sample information may be sample information obtained after data dimension reduction processing. Specifically, data dimension reduction processing may be performed on the preset sample information based on principal component analysis technology to obtain the processed sample information.
[0033] Optionally, principal component analysis is performed on the preset sample information to obtain processed sample information, including: determining a covariance matrix corresponding to the preset sample information; performing eigenvalue decomposition on the covariance matrix to obtain eigenvalues and eigenvectors of the covariance matrix; determining a score matrix based on the eigenvalues and eigenvectors, and determining the processed sample information based on the score matrix.
[0034] Among them, the preset sample information can be standardized to obtain the data matrix X, and then the covariance matrix corresponding to the data matrix X can be determined. Specifically, the eigenvalues and eigenvectors corresponding to the covariance matrix can be determined, and then the load matrix can be determined according to the eigenvalues and eigenvectors, and then the score matrix can be determined according to the load matrix and the data matrix X, and finally the score matrix and the preset sample information can be used together as the processed sample information. Exemplarily, the steps of determining the score matrix are as follows:
[0035] Define the covariance matrix of the historical data matrix X after standardization as S = cov(x);
[0036] ② Perform eigenvalue decomposition on the covariance matrix S to obtain the eigenvalues and eigenvectors of the matrix S, and decompose the data matrix X into the following:
[0037]
[0038] T=XP
[0039] Where P∈R I×A is the loading matrix, each column of which is composed of the first A eigenvectors of the covariance matrix S, T∈R N×A It is a score matrix, and each column in the score matrix T is called a principal variable.
[0040] After principal component analysis, each principal component in the score matrix is statistically linearly independent of any other principal component. As the number of principal components increases, the principal component variance gradually decreases, and the ability to explain the data gradually weakens. Selecting the number of principal components with a higher cumulative variance contribution rate can retain as much data information as possible, while avoiding the computational burden and overfitting caused by too many dimensions. Generally, the number of principal components with a cumulative variance contribution rate exceeding the predetermined threshold of 85% is selected as the final dimensionality reduction result.
[0041] S230: Training the preliminary risk detection model based on the processed sample information to obtain the target risk detection model.
[0042] The target risk detection model may be a model that is ultimately used to detect the credit risk of the target object. The target risk detection model may be obtained by training the preliminary risk detection model based on the preset sample information. Specifically, the processed sample information may be input into the preliminary risk detection model to obtain the sample risk detection result; the detection loss function may be determined based on the sample risk detection result and the preset risk label of the processed sample information; the preliminary risk detection model may be adjusted based on the detection loss function to obtain the target risk detection model.
[0043] The sample risk detection result may be a risk detection result of the preliminary risk detection model on the object in the preset sample information. The preset risk label may be a risk type label pre-marked in the preset sample information. Exemplarily, the preset risk label may include a risk-free label and a risky label. By comparing the sample risk detection result with the preset risk label of the preset sample information, the error of the preliminary risk detection model in performing credit risk detection may be known.
[0044] Furthermore, the detection loss function can be determined according to the comparison result. The detection loss function can be a loss function corresponding to the detection error. The error size can be known through the detection loss function. Subsequently, the preliminary risk detection model can be adjusted according to the detection loss function to obtain the target risk detection model. Specifically, the model parameters in the preliminary risk detection model can be adjusted according to the detection loss function to obtain the target risk detection model.
[0045] Optionally, the preliminary risk detection model includes a risk reference determination network and a risk analysis network. The processed sample information is input into the preliminary risk detection model to obtain a sample risk detection result, including: inputting the processed sample information into the risk reference determination network to obtain a target risk threshold result; inputting the target risk threshold result and the processed sample information into the risk analysis network to obtain a sample risk detection result.
[0046] Among them, the risk reference determination network is used to determine the reference standard of credit risk. The risk analysis network can perform credit risk detection on the input credit data based on the reference standard. The target risk threshold result can be a reference standard for determining whether the processed sample information has credit risk. Further, the target risk threshold result and the processed sample information can be input into the risk analysis network, so that the network analyzes whether the processed sample information has risk based on the target risk threshold result, and then obtains the sample risk detection result.
[0047] Optionally, the risk reference determination network includes: a principal component determination subnetwork, a principal component analysis subnetwork and a reference determination subnetwork. The processed sample information is input into the risk reference determination network to obtain a target risk threshold result, including: inputting the processed sample information into the principal component determination subnetwork to obtain multiple candidate principal component variables; inputting every two candidate principal component variables into the principal component analysis subnetwork respectively to determine the target principal component variable pair; inputting the target principal component variable pair into the reference determination subnetwork to determine the target segmentation hyperplane, and using the target segmentation hyperplane as the target risk threshold result.
[0048] The candidate principal variable may be a principal variable that may be used to determine the risk reference standard. Specifically, each column of data in the score matrix may be used as a candidate principal variable. The target principal variable pair may be a pair of principal variables used to determine the risk reference standard. The target principal variable pair may then be input into the reference determination subnetwork so that the network determines the optimal segmentation hyperplane of the target principal variable pair, and the determined optimal segmentation hyperplane is used as the target risk threshold result.
[0049] Optionally, every two candidate principal component variables are input into the principal component analysis sub-network respectively to determine the target principal component variable pair, including: taking every two candidate principal component variables as a candidate principal component variable pair, inputting all candidate principal component variable pairs into the principal component analysis sub-network, and outputting the candidate splitting hyperplanes of each candidate principal component variable pair respectively; and taking the candidate principal component variable pair corresponding to the candidate splitting hyperplane with the largest distance as the target principal component variable pair.
[0050] The candidate segmentation hyperplane may be a segmentation hyperplane corresponding to the candidate principal variable pair. Exemplarily, the formula for determining the candidate segmentation hyperplane of the candidate principal variable pair is as follows:
[0051] ω T x i +b≥1,y i =1
[0052] ω T x i +b≤-1,y i =-1
[0053] Where ω is the normal vector of the hyperplane, b is the bias term, and x i is a candidate principal variable. 1i ,x 2i ) to the optimal separating hyperplane, the greater the distance, the stronger the generalization ability of the support vector machine model and the more accurate the discrimination result.
[0054] Furthermore, the formula for determining the optimal segmentation hyperplane for the target principal variable pair is as follows:
[0055]
[0056] st1-y i (ω T x i +b)≤0
[0057] The optimal segmentation hyperplane can be used to determine whether a credit transaction is a normal transaction or a risky transaction. For example, the credit risk detection result can be determined by the positional relationship between the feature points corresponding to the credit information and the optimal segmentation hyperplane. For example, if the feature points corresponding to the credit information are on one side of the optimal segmentation hyperplane, it is a normal transaction, and on the other side, it is a risky transaction.
[0058] S240: Obtain the credit information of the target object to be analyzed.
[0059] The target object may be an object that needs to be subjected to credit risk detection. The credit information to be analyzed may be credit information that needs to be subjected to credit risk analysis. Specifically, the historical credit information of the target object may be obtained based on a preset acquisition method, and the historical credit information of the target object may be used as the credit information to be analyzed.
[0060] S250: input the credit information to be analyzed into a pre-trained target risk detection model to obtain a target risk detection result of the target object.
[0061] The target risk detection result may be a credit risk detection result of the target object. Exemplarily, the target risk detection result may include detection results of types such as risk-free and risky. Specifically, the credit information to be analyzed may be input into a pre-trained target risk detection model so that the model analyzes the credit information to be analyzed, thereby obtaining the target risk detection result of the target object.
[0062] The technical solution provided by the embodiment of the present invention has the following advantages:
[0063] ① Efficient processing of high-dimensional data: The principal component analysis algorithm maps high-dimensional data to a low-dimensional subspace, retains the maximum variance of the data, reduces redundant information input into the support vector machine, and reduces the risk of overfitting.
[0064] ② Eliminate the collinearity problem between variables: Traditional anomaly discrimination methods are difficult to deal with collinearity problems. This method uses the principal component analysis algorithm to convert the original variables into a set of linearly independent new variables through linear transformation, thereby removing the collinearity in customer credit data and better capturing the potential patterns and laws in the data.
[0065] ③ Fully explore data features: The principal component analysis algorithm retains data information by maximizing data variance when reducing dimensions, while the support vector machine can retain data category information. The combination of these two methods can retain category information while efficiently reducing dimensions, and more comprehensively consider data features to improve the performance and generalization ability of the discriminant model.
[0066] In summary, through the method of combining principal component analysis and support vector machine, the present invention can efficiently process high-dimensional data, reduce model complexity, deal with collinearity problems, fully mine data features and achieve rapid response in credit risk management, thereby effectively solving the challenges faced by existing methods and having significant advantages.
[0067] For example, in order to better understand the technical solution provided by the present invention, a specific embodiment is introduced below:
[0068] Step 1: Integrate the credit data of users who have completed credit transactions into a data set X(I×N) and perform data preprocessing;
[0069] Step 2: Use the principal component analysis algorithm to train and obtain the data feature matrix in the credit transaction scenario, that is, the score matrix T = XP∈R N×A Finally, the variance calculation outputs the two data features that best reflect the trading rules t1 = {t 11 ,t 12 …,t 1A}、t2={t 21 ,t 22 …,t 2A}, the two extracted data features are used as input to establish a credit risk discrimination model and used to establish a risk anomaly discrimination module;
[0070] Step 3: Take the data features output from step 1 and the corresponding credit transaction labels as input, and use the support vector machine algorithm for training to obtain a separating hyperplane that distinguishes whether the credit transaction is normal or abnormal.
[0071] Step 4: Input the real-time customer credit data into the trained discrimination method model, and the model will determine whether the credit transaction is a normal transaction or a risky transaction.
[0072] The technical solution provided by the embodiment of the present invention obtains a preliminary risk detection model and preset sample information; performs principal component analysis on the preset sample information to obtain processed sample information; trains the preliminary risk detection model based on the processed sample information to obtain the target risk detection model; obtains the credit information to be analyzed of the target object; inputs the credit information to be analyzed into the pre-trained target risk detection model to obtain the target risk detection result of the target object. The technical solution of the embodiment of the present invention solves the problems of complex sample data and high computational complexity of model training in existing credit risk analysis technologies. The principal component analysis technology can be used to reduce the dimensionality of high-dimensional sample data, and model training can be performed based on the reduced-dimensional sample data, thereby reducing the computational complexity of model training and improving the efficiency of model training.
[0073] Figure 3It is a structural schematic diagram of a credit risk detection device provided by an embodiment of the present invention. The embodiment of the present invention can be applied to the scenario of detecting the credit risk of a target object. The device can be implemented by software and / or hardware and integrated into a computer device with application development function.
[0074] like Figure 3 As shown, the credit risk detection device includes: a credit information acquisition module 310 and a credit information analysis module 320.
[0075] Among them, the credit information acquisition module 310 is used to obtain the credit information to be analyzed of the target object; the credit information analysis module 320 is used to input the credit information to be analyzed into a pre-trained target risk detection model to obtain the target risk detection result of the target object; wherein, the target risk detection model is trained based on preset sample information after principal component analysis.
[0076] The technical solution provided by the embodiment of the present invention obtains the credit information to be analyzed of the target object; inputs the credit information to be analyzed into a pre-trained target risk detection model to obtain the target risk detection result of the target object; wherein the target risk detection model is trained based on preset sample information after principal component analysis. The technical solution of the embodiment of the present invention solves the problems of complex sample data and high computational complexity of model training in existing credit risk analysis technologies. The principal component analysis technology can be used to reduce the dimensionality of high-dimensional sample data, and model training can be performed based on the reduced-dimensional sample data, thereby reducing the computational complexity of model training and improving the efficiency of model training.
[0077] In an optional embodiment, the credit risk detection device also includes: a model training module, which is used to: obtain a preliminary risk detection model and preset sample information; perform principal component analysis on the preset sample information to obtain processed sample information; and train the preliminary risk detection model based on the processed sample information to obtain the target risk detection model.
[0078] In an optional embodiment, the model training module includes: a principal component analysis unit, used to: determine the covariance matrix corresponding to the preset sample information; perform eigenvalue decomposition on the covariance matrix to obtain the eigenvalues and eigenvectors of the covariance matrix; determine a score matrix based on the eigenvalues and eigenvectors, and determine the processed sample information based on the score matrix.
[0079] In an optional embodiment, the model training module includes: a model adjustment unit, used to: input the processed sample information into the preliminary risk detection model to obtain a sample risk detection result; determine a detection loss function based on the sample risk detection result and a preset risk label of the processed sample information; adjust the preliminary risk detection model according to the detection loss function to obtain the target risk detection model.
[0080] In an optional embodiment, the preliminary risk detection model includes a risk reference determination network and a risk analysis network. In an optional embodiment, the model training module includes: a sample risk detection unit, used to: input the processed sample information into the risk reference determination network to obtain a target risk threshold result; input the target risk threshold result and the processed sample information into the risk analysis network to obtain the sample risk detection result.
[0081] In an optional embodiment, the risk reference determination network includes: a principal component determination subnetwork, a principal component analysis subnetwork and a reference determination subnetwork, and the sample risk detection unit includes: a risk threshold determination subunit, which is used to: input the processed sample information into the principal component determination subnetwork to obtain multiple candidate principal component variables; input every two candidate principal component variables into the principal component analysis subnetwork respectively to determine a target principal component variable pair; input the target principal component variable pair into the reference determination subnetwork to determine a target segmentation hyperplane, and use the target segmentation hyperplane as the target risk threshold result.
[0082] In an optional embodiment, the sample risk detection unit includes: a target principal component variable pair determination subunit, which is used to: take every two candidate principal component variables as a candidate principal component variable pair, input all candidate principal component variable pairs into the principal component analysis subnetwork, and output candidate splitting hyperplanes for each candidate principal component variable pair; and take the candidate principal component variable pair corresponding to the candidate splitting hyperplane with the largest distance as the target principal component variable pair.
[0083] The credit risk detection device provided in the embodiment of the present invention can execute the credit risk detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.
[0084] Figure 4 A schematic diagram of the structure of a computer device provided in an embodiment of the present invention. Figure 4 A block diagram of an exemplary computer device 12 suitable for use in implementing embodiments of the present invention is shown. Figure 4The computer device 12 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention. The computer device 12 can be any terminal device with computing capability and can be configured in the credit risk detection device.
[0085] like Figure 4 As shown, the computer device 12 is in the form of a general-purpose computing device. The components of the computer device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that connects various system components (including the system memory 28 and the processing unit 16).
[0086] The bus 18 may be one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. For example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.
[0087] The computer device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the computer device 12, including volatile and non-volatile media, removable and non-removable media.
[0088] The system memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache 32. The computer device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 4 not shown, usually called a "hard drive"). Although Figure 4 Not shown in the figure, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The system memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present invention.
[0089] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in system memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. Program modules 42 generally perform the functions and / or methods of the embodiments described herein.
[0090] The computer device 12 may also communicate with one or more external devices 14 (e.g., keyboards, pointing devices, displays 24, etc.), one or more devices that enable a user to interact with the computer device 12, and / or any device that enables the computer device 12 to communicate with one or more other computing devices (e.g., network cards, modems, etc.). Such communication may be performed through an input / output (I / O) interface 22. Furthermore, the computer device 12 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) through a network adapter 20. Figure 4 As shown, the network adapter 20 communicates with other modules of the computer device 12 via the bus 18. It should be understood that although Figure 4 Not shown, other hardware and / or software modules may be used in conjunction with computer device 12, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.
[0091] The processing unit 16 executes various functional applications and data processing by running the programs stored in the system memory 28, for example, implementing the credit risk detection method provided in the embodiment of the present invention, which includes:
[0092] Obtain the target object's credit information to be analyzed; input the credit information to be analyzed into a pre-trained target risk detection model to obtain a target risk detection result of the target object; wherein the target risk detection model is trained based on preset sample information after principal component analysis.
[0093] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the credit risk detection method provided by any embodiment of the present invention is implemented, including:
[0094] Obtain the target object's credit information to be analyzed; input the credit information to be analyzed into a pre-trained target risk detection model to obtain a target risk detection result of the target object; wherein the target risk detection model is trained based on preset sample information after principal component analysis.
[0095] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with 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), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.
[0096] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0097] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
[0098] Computer program code for performing the operations of the present invention may be written in one or more programming languages or combinations thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).
[0099] It should be understood by those skilled in the art that the modules or steps of the present invention described above can be implemented by a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, optionally, they can be implemented by a program code executable by a computer device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.
[0100] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims
1. A credit risk detection method, characterized in that: include: Obtain the target object's credit information to be analyzed; Inputting the credit information to be analyzed into a pre-trained target risk detection model to obtain a target risk detection result of the target object; The target risk detection model is obtained by training based on preset sample information after principal component analysis.
2. The method according to claim 1, characterized in that The training process of the target risk detection model includes: Obtain preliminary risk detection models and preset sample information; Performing principal component analysis on the preset sample information to obtain processed sample information; The preliminary risk detection model is trained based on the processed sample information to obtain the target risk detection model.
3. The method according to claim 2, characterized in that The performing principal component analysis on the preset sample information to obtain processed sample information includes: Determine the covariance matrix corresponding to the preset sample information; Performing eigenvalue decomposition on the covariance matrix to obtain eigenvalues and eigenvectors of the covariance matrix; A score matrix is determined according to the eigenvalues and the eigenvectors, and the processed sample information is determined according to the score matrix.
4. The method according to claim 3, characterized in that The training of the preliminary risk detection model based on the processed sample information to obtain the target risk detection model includes: Inputting the processed sample information into the preliminary risk detection model to obtain a sample risk detection result; Determine a detection loss function according to the sample risk detection result and a preset risk label of the processed sample information; The preliminary risk detection model is adjusted according to the detection loss function to obtain the target risk detection model.
5. The method according to claim 4, characterized in that The preliminary risk detection model includes a risk reference determination network and a risk analysis network. The processed sample information is input into the preliminary risk detection model to obtain a sample risk detection result, including: Inputting the processed sample information into the risk reference determination network to obtain a target risk threshold result; The target risk threshold result and the processed sample information are input into the risk analysis network to obtain the sample risk detection result.
6. The method according to claim 5, characterized in that The risk reference determination network includes: a principal component determination subnetwork, a principal component analysis subnetwork and a reference determination subnetwork. The inputting of the processed sample information into the risk reference determination network to obtain a target risk threshold result includes: Inputting the processed sample information into the principal component determination subnetwork to obtain a plurality of candidate principal component variables; Inputting each two candidate principal component variables into the principal component analysis subnetwork respectively to determine a target principal component variable pair; The target principal component variable pair is input into the reference determination subnetwork to determine the target segmentation hyperplane, and the target segmentation hyperplane is used as the target risk threshold result.
7. The method according to claim 6, characterized in that The step of inputting each two candidate principal component variables into the principal component analysis subnetwork to determine a target principal component variable pair comprises: Each two candidate principal component variables are respectively regarded as a candidate principal component variable pair, all candidate principal component variable pairs are input into the principal component analysis subnetwork, and candidate segmentation hyperplanes of each candidate principal component variable pair are respectively output; The candidate principal component variable pair corresponding to the candidate segmentation hyperplane with the largest distance is used as the target principal component variable pair.
8. A credit risk detection device, characterized in that: The device comprises: A credit information acquisition module is used to obtain the credit information of the target object to be analyzed; The credit information analysis module is used to input the credit information to be analyzed into a pre-trained target risk detection model to obtain the target risk detection result of the target object; wherein the target risk detection model is trained based on preset sample information after principal component analysis.
9. A computer device, characterized in that: The computer device comprises: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the credit risk detection method as described in any one of claims 1-7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the credit risk detection method as described in any one of claims 1 to 7 is implemented.