Credit risk prediction method, electronic device and readable storage medium

Abstract

The application discloses a credit risk prediction method, an electronic device and a readable storage medium, and is applied to the technical field of financial technology. The credit risk prediction method comprises the following steps: acquiring a historical behavior data set of a user, wherein the historical behavior data set is derived from at least two data sources; integrating the historical behavior data set into historical behavior fusion data; and performing credit risk prediction on the user according to the historical behavior fusion data and a risk prediction model, to obtain a risk prediction result. The application solves the technical problem that the prediction accuracy and the prediction efficiency of user credit risk cannot be considered simultaneously.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence technology in financial technology (Fintech), and in particular to a credit risk prediction method, an electronic device and a readable storage medium. BACKGROUND

[0002] With the continuous development of financial technology, especially Internet technology finance, more and more technologies (such as distributed, artificial intelligence, etc.) are applied in the financial field, but the financial industry also puts forward higher requirements for technology, such as higher requirements for the credit level of users in the financial industry.

[0003] At present, in order to evaluate the credit level of users, the credit risk of users is usually predicted according to the historical behavior data set of users and the risk prediction model. In order to improve the prediction accuracy of the credit risk of users, the historical behavior data set is usually derived from multiple channels and multiple data sources, and when the model processes a large amount of multi-dimensional historical behavior data set, due to the high processing complexity, it is easy to take a long time to process or even cannot be processed, therefore, the prediction accuracy and the prediction efficiency of the credit risk of users cannot be taken into account. SUMMARY

[0004] The main purpose of the present application is to provide a credit risk prediction method, an electronic device and a readable storage medium, which aims to solve the technical problem that the prediction accuracy and the prediction efficiency of the credit risk of users cannot be taken into account in the prior art.

[0005] To achieve the above-mentioned purpose, the present application provides a credit risk prediction method, which comprises:

[0006] obtaining a historical behavior data set of a user, wherein the historical behavior data set is derived from at least two data sources;

[0007] integrating the historical behavior data set into historical behavior fusion data;

[0008] predicting the credit risk of the user according to the historical behavior fusion data and a risk prediction model, to obtain a risk prediction result.

[0009] Optionally, the step of integrating the historical behavior data set into historical behavior fusion data comprises:

[0010] determining a variable fusion coefficient corresponding to each variable in the historical behavior data set derived from each data source;

[0011] integrating each variable into historical behavior fusion data according to each variable fusion coefficient.

[0012] Optionally, the step of determining the variable fusion coefficients corresponding to each variable in the historical behavior data set from each data source comprises:

[0013] decomposing the historical behavior data set to obtain a decomposition matrix;

[0014] determining the variable fusion coefficients corresponding to each variable in the historical behavior data set from each data source according to the historical behavior data set and the decomposition matrix.

[0015] Optionally, the decomposition matrix comprises a result matrix and a coefficient matrix,

[0016] the step of decomposing the historical behavior data set to obtain a decomposition matrix comprises:

[0017] decomposing the historical behavior data set by a preset matrix decomposition method to obtain a result matrix and a coefficient matrix, wherein the result matrix and the coefficient matrix are non-negative matrices.

[0018] Optionally, the decomposition matrix comprises a result matrix,

[0019] the step of determining the variable fusion coefficients corresponding to each variable in the historical behavior data set from each data source according to the historical behavior data set and the decomposition matrix comprises:

[0020] taking the historical behavior data set as a regression model input value and taking the result matrix as an input label corresponding to the regression model input value;

[0021] fitting the variable fusion coefficients corresponding to each variable in the historical behavior data set from each data source by a multiple linear regression model according to the regression model input value and the input label.

[0022] Optionally, the step of integrating each variable into historical behavior fusion data according to each variable fusion coefficient comprises:

[0023] determining the product between each variable and each corresponding variable fusion coefficient;

[0024] integrating each product into historical behavior fusion data.

[0025] Optionally, before the step of predicting the credit risk of the user according to the historical behavior fusion data and a risk prediction model to obtain a risk prediction result, the method further comprises:

[0026] obtaining a training behavior data set and a real label corresponding to the training behavior data set, wherein the training behavior data set is derived from at least two data sources;

[0027] integrating the training behavior data set into training behavior fusion data;

[0028] iteratively optimizing a risk prediction model to be trained according to the training behavior fusion data and the true label, to obtain the risk prediction model.

[0029] To achieve the above object, the present application further provides a credit risk prediction device, which comprises:

[0030] an acquisition module configured to acquire a historical behavior data set of a user, wherein the historical behavior data set is derived from at least two data sources;

[0031] an integration module configured to integrate the historical behavior data set into historical behavior fusion data;

[0032] a prediction module configured to perform credit risk prediction on the user according to the historical behavior fusion data and a risk prediction model, to obtain a risk prediction result.

[0033] Optionally, the integration module is further configured to:

[0034] determine a variable fusion coefficient corresponding to each variable derived from each data source in the historical behavior data set;

[0035] integrate each variable into historical behavior fusion data according to each variable fusion coefficient.

[0036] Optionally, the integration module is further configured to:

[0037] perform decomposition on the historical behavior data set to obtain a decomposition matrix;

[0038] determine a variable fusion coefficient corresponding to each variable derived from each data source in the historical behavior data set according to the historical behavior data set and the decomposition matrix.

[0039] Optionally, the decomposition matrix comprises a result matrix and a coefficient matrix, and the integration module is further configured to:

[0040] perform decomposition on the historical behavior data set by a preset matrix decomposition method to obtain a result matrix and a coefficient matrix, wherein the result matrix and the coefficient matrix are non-negative matrices.

[0041] Optionally, the decomposition matrix comprises a result matrix, and the integration module is further configured to:

[0042] regard the historical behavior data set as input values of a regression model, and regard the result matrix as input labels corresponding to the input values of the regression model;

[0043] According to the regression model input value and the input label, a variable fusion coefficient corresponding to each variable in the historical behavior data set from each data source is obtained by a multiple linear regression model fitting.

[0044] Optionally, the integration module is further configured to:

[0045] determine a product between each variable and each corresponding variable fusion coefficient;

[0046] integrate each product into historical behavior fusion data.

[0047] Optionally, before the step of predicting credit risk of the user according to the historical behavior fusion data and a risk prediction model to obtain a risk prediction result, the credit risk prediction device is further configured to:

[0048] obtain a training behavior data set and a true label corresponding to the training behavior data set, wherein the training behavior data set is at least from two data sources;

[0049] integrate the training behavior data set into training behavior fusion data;

[0050] According to the training behavior fusion data and the true label, a risk prediction model to be trained is iteratively optimized to obtain a risk prediction model.

[0051] The present application also provides an electronic device, comprising a memory, a processor, and a program of the credit risk prediction method stored in the memory and executable on the processor, wherein the program of the credit risk prediction method, when executed by the processor, can implement the steps of the credit risk prediction method as described above.

[0052] The present application also provides a computer readable storage medium, wherein a program of the credit risk prediction method is stored on the computer readable storage medium, and the program of the credit risk prediction method, when executed by the processor, implements the steps of the credit risk prediction method as described above.

[0053] The present application also provides a computer program product, comprising a computer program, wherein the computer program, when executed by the processor, implements the steps of the credit risk prediction method as described above.

[0054] The application provides a credit risk prediction method, an electronic device and a readable storage medium. Compared with a method for predicting credit risks of a user according to a historical behavior data set of the user and a risk prediction model, the application obtains the historical behavior data set of the user, wherein the historical behavior data set is derived from at least two data sources; integrates the historical behavior data set into historical behavior fusion data; and predicts the credit risk of the user according to the historical behavior fusion data and the risk prediction model, to obtain a risk prediction result. The historical behavior data set derived from at least two data sources, i.e., a large amount of multi-dimensional historical behavior data set, is integrated into the historical behavior fusion data. The historical behavior fusion data is fusion data containing historical behavior data set information, and the historical behavior fusion data is input data of the risk prediction model. Therefore, the complexity of the input data of the risk prediction model is reduced, the prediction accuracy of the risk prediction model is improved, the complexity of model processing is reduced, the technical defect that the model directly processes a large amount of multi-dimensional historical behavior data set and is prone to a situation of long processing time or even unable to process is avoided, the prediction efficiency of the credit risk of the user is improved, and the prediction efficiency of the credit risk of the user is improved while ensuring high prediction accuracy of the credit risk of the user. BRIEF DESCRIPTION OF DRAWINGS

[0055] The accompanying drawings, which are incorporated into and form a part of the specification, illustrate an embodiment consistent with the present application and, together with the description, serve to explain the principles of the application.

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, for those skilled in the art, other drawings can also be obtained based on these drawings without creative labor.

[0057] Figure 1 A flowchart of a first embodiment of the credit risk prediction method of the present application;

[0058] Figure 2 A matrix decomposition diagram of a historical behavior data set in a scenario related to the credit risk prediction method of the present application;

[0059] Figure 3 A matrix decomposition diagram of a historical behavior data set in another scenario related to the credit risk prediction method of the present application;

[0060] Figure 4 A variable fusion coefficient fitting method in a scenario related to the credit risk prediction method of the present application;

[0061] Figure 5A device structure schematic diagram related to the credit risk prediction method of the present application;

[0062] Figure 6 A device structure schematic diagram of a hardware running environment related to the credit risk prediction method in the embodiments of the present application.

[0063] The purposes, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0064] In order to make the above purposes, features and advantages of the present application more apparent and easy to understand, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0065] Embodiment one

[0066] The credit risk prediction method provided in the embodiments of the present application is described below. Figure 1 The credit risk prediction method comprises the following steps.

[0067] Step S10, obtaining a historical behavior data set of a user, wherein the historical behavior data set is derived from at least two data sources;

[0068] In the present embodiment, it should be noted that the number of users can be one or multiple, the historical behavior data is the behavior data of the user at the last time step, the time step can be set as a periodically extracted time period, the number of historical behavior data is multiple, and the data source can be a bank source, a social software source, a travel software source or other sources.

[0069] Exemplarily, step S10 comprises: obtaining the behavior data of the user at the last time step from at least two data sources to obtain a historical behavior data set, wherein the time step can be one month, one year or three years; the time step can be the best evaluation time experience value for evaluating the credit risk behavior of the user, or can be set by the user credit risk evaluation party; the historical behavior data can be the behavior data of multiple time periods of the same user, the behavior data of multiple time periods of different users, or the behavior data of the same time period of different users.

[0070] Step S20, integrating the historical behavior data set into historical behavior fusion data;

[0071] As an example, step S20 includes: extracting features from the historical behavior dataset to obtain feature information from each of the data sources; fusing the feature information to obtain fused feature information; and using the fused feature information as the historical behavior fused data.

[0072] As an example, step S20 includes: obtaining the data relationships between the data in the historical behavior dataset, and integrating the historical dataset into historical behavior fusion data according to the data relationships, wherein the data relationships can be determined by the data type corresponding to each of the data, or by the data attribute corresponding to each of the data, or by the data feature corresponding to each of the data.

[0073] As an example, step S20 includes: performing dimensionality reduction on the historical behavior dataset to obtain a dimensionality reduction matrix, and integrating the historical behavior dataset into historical behavior fusion data based on the dimensionality reduction matrix.

[0074] Step S30: Based on the historical behavior fusion data and risk prediction model, perform credit risk prediction on the user to obtain the risk prediction result.

[0075] In this embodiment, it should be noted that the risk prediction model can be a scorecard model, a tree model, or a deep learning model.

[0076] For example, step S30 includes: extracting features from the historical behavior fusion data to obtain historical behavior fusion features; and mapping the historical behavior fusion features to the risk prediction result corresponding to the user through a risk prediction model.

[0077] In step S20, the step of integrating the historical behavior dataset into historical behavior fusion data includes:

[0078] Step S21: Determine the variable fusion coefficients corresponding to each variable from each of the data sources in the historical behavior dataset;

[0079] It is understandable that there may be a linear relationship between the same variable from different data sources. Therefore, as an example, step S21 includes: obtaining the linear relationship between the variables from each of the data sources in the historical behavior dataset; and generating the variable fusion coefficients corresponding to each variable from each of the data sources in the historical behavior dataset based on the linear relationship.

[0080] As an example, step S21 includes: obtaining the pairwise Euclidean distances between each variable in the historical behavior dataset originating from each of the data sources, and generating variable fusion coefficients corresponding to each variable in the historical behavior dataset originating from each of the data sources based on each of the Euclidean distances.

[0081] As an example, step S21 includes: obtaining the cosine distance between each pair of variables from each of the data sources in the historical behavior dataset, and generating variable fusion coefficients corresponding to each variable from each of the data sources in the historical behavior dataset based on each cosine distance.

[0082] Step S22: Based on the fusion coefficients of each variable, integrate the variables into historical behavior fusion data.

[0083] For example, step S22 includes: integrating the variables into historical behavior fusion data by using a preset fusion algorithm based on the fusion coefficients of each variable.

[0084] By using the fusion coefficients of variables from different data sources in the historical behavior dataset to reduce the dimensionality of the historical behavior dataset, a balance is achieved between preserving the linear relationships between variables from different data sources and reducing the dimensionality of the historical behavior dataset. This reduces the complexity of model processing and preserves the richness of the historical behavior dataset. Since the historical behavior fusion data is the input data for the risk prediction model, it reduces the complexity of the input data for the risk prediction model and improves the prediction accuracy of the risk prediction model.

[0085] In step S21, the step of determining the variable fusion coefficients corresponding to each variable from each of the data sources in the historical behavior dataset includes:

[0086] Step A10: Decompose the historical behavior dataset to obtain a decomposition matrix;

[0087] Step A20: Based on the historical behavior dataset and the decomposition matrix, determine the variable fusion coefficients corresponding to each variable in the historical behavior dataset originating from each of the data sources.

[0088] As an example, steps A10 to A20 include: performing eigenvalue decomposition on the historical behavior dataset to obtain a decomposition matrix; and determining the variable fusion coefficients corresponding to each variable from each of the data sources in the historical behavior dataset based on the historical behavior dataset and the decomposition matrix.

[0089] In step A10, the decomposition matrix includes a result matrix and a coefficient matrix, and the step of decomposing the historical behavior data set to obtain a decomposition matrix includes:

[0090] In step A11, the historical behavior data set is decomposed by a preset matrix decomposition method to obtain a result matrix and a coefficient matrix, wherein the result matrix and the coefficient matrix are non-negative matrices.

[0091] It can be understood that when the above matrix decomposition method is used, since the same variable from different data sources may have a collinearity relationship with each other or a negative correlation with each other, and different linear relationships of the same variable have different explanatory properties for the model (for example, the linear coefficient of variable X from data source 1 is positive, that is, it has the property of risk interpretation for the model, the linear coefficient of variable X from data source 2 is negative, that is, it has the property of not meeting the risk interpretation for the model, the linear coefficient of variable X from data source 3 is 0, that is, it has the property of no effect for the model), it is easy to appear that the contribution of the variable from part of the data sources in the historical behavior data set to the prediction result of the model is low or has no contribution, or even has a reverse effect on the prediction result of the model, thereby resulting in low prediction accuracy of the user credit risk.

[0092] To overcome the above defects, as an example, step A11 includes: obtaining a preset matrix decomposition method, and decomposing the historical behavior data set into a result matrix of a preset data dimension and a one-dimensional coefficient matrix by the preset matrix decomposition method, wherein the preset data dimension is consistent with the number of data sources from which the historical behavior data set is derived, and the preset matrix decomposition method is a non-negative matrix factorization (Non-negative Matrix Factorization, NMF) algorithm.

[0093] As an example, refer to Figure 2 , Figure 2 A matrix decomposition diagram of a historical behavior data set in a scenario related to the credit risk prediction method of the present application, Figure 2 includes a non-negative historical behavior data set (the matrix V shown in the figure), a non-negative result matrix (the matrix W shown in the figure), and a non-negative coefficient matrix (the matrix H shown in the figure). The historical behavior data set of a preset data quantity (m shown in the figure) and a preset data dimension (n shown in the figure) is decomposed into a result matrix of a dimension of a preset data quantity and a data quantity corresponding to a target dimension (r shown in the figure) and a coefficient matrix of a data quantity corresponding to a preset data dimension and a target dimension.

[0094] As an example, refer to Figure 3 , Figure 3A matrix decomposition schematic diagram of a historical behavior data set in another scenario involved in the credit risk prediction method of the present application, Figure 3 Comprise: historical behavior data set (illustrated x_1_1, x_2_1, x_3_1,..., x_n_1) of data quantity M, result matrix (illustrated vertically placed matrix) and coefficient matrix (illustrated horizontally placed matrix). The historical behavior data of data quantity M is decomposed into result matrix of dimension M, data quantity 1 and 1-dimensional data quantity N matrix by pre-setting matrix decomposition method.

[0095] Wherein, in step A20, the decomposition matrix includes the result matrix, and the step of determining the variable fusion coefficient corresponding to each variable from each data source in the historical behavior data set according to the historical behavior data set and the decomposition matrix comprises:

[0096] Step A21, the historical behavior data set is taken as the regression model input value, and the result matrix is taken as the input label corresponding to the regression model input value.

[0097] Step A22, according to the regression model input value and the input label, the variable fusion coefficient corresponding to each variable from each data source in the historical behavior data set is obtained by fitting the multiple linear regression model.

[0098] Exemplarily, steps A21 to A22 include: taking the historical behavior data set as the regression model input value, that is, input x, and taking the result matrix as the input label corresponding to the regression model input value, that is, input label y; by substituting the input x and the input label y into the multiple linear regression model, the variable fusion coefficient corresponding to each variable from each data source in the historical behavior data set is obtained.

[0099] As an example, refer to Figure 4 , Figure 4 The variable fusion coefficient fitting method in one scenario involved in the credit risk prediction method of the present application, Figure 4 Comprise: historical behavior data set (illustrated matrix V composed of x_1_1, x_2_1, x_3_1,..., x_n_1) of data quantity M, result matrix (illustrated M×1 matrix W) and variable fusion coefficient (illustrated N×1 matrix) corresponding to each variable from each data source, by taking historical behavior data set as X, taking result matrix as Y, and solving the variable fusion coefficient corresponding to each variable from each data source by multiple regression model.

[0100] Compared with the method of predicting the credit risk of a user according to a historical behavior data set of the user and a risk prediction model, the embodiment of the application integrates a large amount of multi-dimensional historical behavior data sets into historical behavior fusion data, thereby reducing the complexity of model processing, avoiding the technical defects that the model takes a long time to process or even cannot process the large amount of multi-dimensional historical behavior data sets due to high processing complexity, and improving the prediction efficiency of the credit risk of the user while ensuring high prediction accuracy of the credit risk of the user.

[0101] Embodiment two

[0102] Further, based on the first embodiment of the application, in another embodiment of the application, the same or similar contents as the above embodiment one can refer to the above introduction, and will not be described in detail. On this basis, in step S22, the step of integrating each variable into historical behavior fusion data according to each variable fusion coefficient includes:

[0103] Step B10, determining the product between each variable and each corresponding variable fusion coefficient;

[0104] Step B20, integrating each product into historical behavior fusion data.

[0105] Exemplarily, steps B10 to B20 include: determining the product between each variable and each corresponding variable fusion coefficient, adding each product to obtain the historical behavior fusion data, for example, the historical behavior data set includes variable X, the fusion coefficient corresponding to the variable X from data source 1 is 0.2, the fusion coefficient corresponding to the variable X from data 2 is 0.4, and the fusion coefficient corresponding to the variable X from data 3 is 0.4, so the historical behavior fusion data is 0.2X+0.4X+0.4X.

[0106] In step S30, before the step of predicting the credit risk of the user according to the historical behavior fusion data and the risk prediction model to obtain the risk prediction result, it further includes:

[0107] Step C10, obtaining a training behavior data set and a real label corresponding to the training behavior data set, wherein the training behavior data set is derived from at least two data sources;

[0108] Step C20, integrating the training behavior data set into training behavior fusion data;

[0109] Step C30, iteratively optimizing the to-be-trained risk prediction model according to the training behavior fusion data and the true label, to obtain a risk prediction model.

[0110] It can be understood that, in the to-be-trained risk prediction model, in order to improve the prediction accuracy of the trained risk prediction model, the training behavior data set used to train the to-be-trained risk prediction model is usually derived from multiple channels and multiple data sources. However, when the model is trained and processed according to a large amount of multi-dimensional historical behavior data set, due to the high processing complexity, it is easy to have a long processing time or even cannot be processed. Therefore, it is difficult to balance the prediction accuracy of the credit risk prediction model and the model training efficiency.

[0111] Exemplarily, steps C10 to C30 include: obtaining a training behavior data set and a true label corresponding to the training behavior data set, wherein the training behavior data set is derived from at least two data sources; integrating the training behavior data set into training behavior fusion data; mapping the training behavior fusion data into a training risk prediction result corresponding to the training behavior data set by a to-be-trained risk prediction model; calculating a model loss corresponding to the to-be-trained risk prediction model according to the difference between the training risk prediction result and the true label, and then judging whether the model loss converges. If the model loss converges, the to-be-trained risk prediction model is used as the risk prediction model. If the model loss does not converge, the to-be-trained risk prediction model is updated by a preset model updating method based on the gradient calculated by the model loss, and the step of obtaining a training behavior data set and a true label corresponding to the training behavior data set is returned, wherein the training behavior data set is derived from at least two data sources, and the preset model updating method includes gradient descent method and gradient ascent method.

[0112] The integration of the training behavior data set into training behavior fusion data can refer to the specific embodiments of steps S20, S21 to S22, A10 to A20, A11 and A21 to A22, which are not repeated here.

[0113] By integrating a large amount of multi-dimensional training behavior data set into training behavior fusion data, the complexity of model training is reduced, thereby avoiding the technical defects that the model is trained and processed according to a large amount of multi-dimensional historical behavior data set, and the processing time is long or even cannot be processed due to high processing complexity. Therefore, the model training efficiency of the credit risk prediction model is improved while ensuring the prediction accuracy of the credit risk prediction model.

[0114] The embodiment of the present application provides a credit risk prediction method, and the embodiment of the present application determines the product between each variable and each corresponding variable fusion coefficient; integrates each product into historical behavior fusion data, thereby reducing the dimension of the historical behavior data set while retaining the data complexity, the historical behavior fusion data is input data of a risk prediction model, so that the complexity of the input data of the risk prediction model is reduced and the prediction accuracy of the risk prediction model is improved.

[0115] Embodiment three

[0116] The embodiment of the present application also provides a credit risk prediction device, referring to Figure 5 , the credit risk prediction device comprises:

[0117] The acquisition module is configured to acquire a historical behavior data set of a user, wherein the historical behavior data set is derived from at least two data sources;

[0118] The integration module is configured to integrate the historical behavior data set into historical behavior fusion data;

[0119] The prediction module is configured to perform credit risk prediction on the user according to the historical behavior fusion data and a risk prediction model, and obtain a risk prediction result.

[0120] Optionally, the integration module is further configured to:

[0121] Determine variable fusion coefficients corresponding to each variable in the historical behavior data set derived from each data source;

[0122] Integrate each variable into historical behavior fusion data according to each variable fusion coefficient.

[0123] Optionally, the integration module is further configured to:

[0124] Decompose the historical behavior data set to obtain a decomposition matrix;

[0125] According to the historical behavior data set and the decomposition matrix, determine variable fusion coefficients corresponding to each variable in the historical behavior data set derived from each data source.

[0126] Optionally, the decomposition matrix comprises a result matrix and a coefficient matrix, and the integration module is further configured to:

[0127] Obtain a result matrix and a coefficient matrix by decomposing the historical behavior data set through a preset matrix decomposition method, wherein the result matrix and the coefficient matrix are non-negative matrices.

[0128] Optionally, the decomposition matrix comprises a result matrix, and the integration module is further configured to:

[0129] inputting the historical behavior data set as regression model input values and the result matrix as input labels corresponding to the regression model input values;

[0130] fitting the historical behavior data set by a multiple linear regression model according to the regression model input values and the input labels to obtain variable fusion coefficients corresponding to each variable from each data source.

[0131] Optionally, the integration module is further configured to:

[0132] determining a product between each variable and each corresponding variable fusion coefficient;

[0133] integrating each product into historical behavior fusion data.

[0134] Optionally, before the step of predicting credit risk of the user according to the historical behavior fusion data and a risk prediction model to obtain a risk prediction result, the credit risk prediction device is further configured to:

[0135] obtaining a training behavior data set and a real label corresponding to the training behavior data set, wherein the training behavior data set is at least from two data sources;

[0136] integrating the training behavior data set into training behavior fusion data;

[0137] iteratively optimizing a to-be-trained risk prediction model according to the training behavior fusion data and the real label to obtain a risk prediction model.

[0138] The credit risk prediction device provided in the application adopts the credit risk prediction method in the above embodiments, and solves the technical problem that the prediction accuracy and prediction efficiency of user credit risk cannot be considered simultaneously. Compared with the prior art, the credit risk prediction device provided in the embodiments has the same beneficial effects as the credit risk prediction method provided in the above embodiments, and other technical features in the credit risk prediction device are the same as the features disclosed in the above embodiments, which will not be repeated here.

[0139] Embodiment Four

[0140] The embodiments of the application provide an electronic device, which comprises at least one processor and a memory in communication connection with the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the credit risk prediction method in the above embodiments.

[0141] The following refers to Figure 6The diagram illustrates a structural schematic of an electronic device suitable for implementing embodiments of the present disclosure. The electronic devices in the embodiments of the present disclosure may include, but are not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Tablet PCs), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0142] like Figure 6 As shown, an electronic device may include a processing unit (such as a central processing unit, graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in ROM (Read-Only Memory) or programs loaded from storage devices into RAM (Random Access Memory). RAM also stores various programs and data required for the operation of the electronic device. The processing unit, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0143] Typically, the following systems can be connected to the I / O interface: input devices including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices including, for example, magnetic tapes, hard disks, etc.; and communication devices. Communication devices allow electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although electronic devices with various systems are shown in the figures, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems may be implemented alternatively.

[0144] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a processing device, it performs the functions defined above in the methods of embodiments of this disclosure.

[0145] The electronic device provided in the present application adopts the credit risk prediction method in the above embodiment, and solves the technical problem that the prediction accuracy and prediction efficiency of user credit risk cannot be considered simultaneously. Compared with the prior art, the electronic device provided in the embodiment of the present application has the same beneficial effects as the credit risk prediction method provided in the above embodiment, and other technical features in the electronic device are the same as the features disclosed in the above embodiment method, which will not be described herein.

[0146] It should be understood that parts of the present disclosure can be realized by hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0147] The above merely provides a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0148] Example Five

[0149] The present embodiment provides a computer-readable storage medium having computer-readable program instructions stored thereon, and the computer-readable program instructions are used to execute the credit risk prediction method in the above embodiment.

[0150] The computer-readable storage medium provided in the embodiments of the present application may, for example, be a U disk, but is not limited to an electric, magnetic, optical, electromagnetic, infrared, or semiconductor system, system, or device, or any combination thereof. More specific examples of the computer-readable storage medium can include, but are not limited to, an electric connection with one or more conductive wires, a portable computer disk, a hard disk, RAM, ROM, EPROM (Erasable Programmable Read Only Memory), or a flash memory, an optical fiber, a CD-ROM (compact disc read-only memory), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present embodiment, the computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium can be transmitted by any suitable medium, including but not limited to an electric wire, an optical cable, an RF (Radio Frequency), etc., or any suitable combination thereof.

[0151] The computer readable storage medium described above can be included in an electronic device; or can exist separately from the electronic device.

[0152] The computer readable storage medium described above carries one or more programs, when the one or more programs are executed by the electronic device, cause the electronic device to: obtain a historical behavior data set of a user, wherein the historical behavior data set is derived from at least two data sources; integrate the historical behavior data set into historical behavior fusion data; and perform credit risk prediction on the user according to the historical behavior fusion data and a risk prediction model, to obtain a risk prediction result.

[0153] Computer program code for carrying out operations of the present disclosure can be written in one or more programming languages or combinations of languages including object oriented programming languages such as Java, Smalltalk, C++ or conventional procedural programming languages such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a LAN or a WAN, or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0154] The flow diagrams and the block diagrams in the drawings are illustrations of possible architectures, functions, and operations for systems, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flow diagrams or block diagrams can represent a module, a segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and / or flow diagrams, and combinations thereof, can be implemented by special purpose hardware-based systems that perform the specified functions or operations, or combinations of special purpose hardware and computer instructions.

[0155] The modules described in the embodiments of the present disclosure can be implemented in the form of software, or can be implemented in the form of hardware. In some cases, the name of the module does not constitute a limitation on the unit itself.

[0156] The computer readable storage medium provided in the present application stores computer readable program instructions for executing the credit risk prediction method described above, and solves the technical problem that the prediction accuracy and prediction efficiency of user credit risk cannot be considered. Compared with the prior art, the beneficial effects of the computer readable storage medium provided in the embodiments of the present application are the same as those of the credit risk prediction method provided in the above embodiments, and are not described here.

[0157] Embodiment six

[0158] The present application also provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the credit risk prediction method as described above.

[0159] The computer program product provided in the present application solves the technical problem that the prediction accuracy and prediction efficiency of user credit risk cannot be considered. Compared with the prior art, the beneficial effects of the computer program product provided in the embodiments of the present application are the same as those of the credit risk prediction method provided in the above embodiments, and are not described here.

[0160] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation using the content of the present application specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent processing scope of the present application.

Claims

1. A credit risk prediction method characterized by, The credit risk prediction method includes: Obtain a user's historical behavior dataset, wherein the historical behavior dataset originates from at least two data sources; The historical behavior dataset is integrated into historical behavior fusion data; Based on the historical behavior fusion data and risk prediction model, the credit risk of the user is predicted, and the risk prediction result is obtained. The step of integrating the historical behavior dataset into historical behavior fusion data includes: Determine the variable fusion coefficients corresponding to each variable originating from each of the aforementioned data sources in the historical behavior dataset; Based on the fusion coefficients of each variable, the variables are integrated into historical behavior fusion data; The step of determining the variable fusion coefficients corresponding to each variable from each of the data sources in the historical behavior dataset includes: The historical behavior dataset is decomposed to obtain a decomposition matrix; Based on the historical behavior dataset and the decomposition matrix, determine the variable fusion coefficients corresponding to each variable in the historical behavior dataset originating from each of the data sources; The decomposition matrix includes a result matrix and a coefficient matrix. The step of decomposing the historical behavior dataset to obtain the decomposition matrix includes: The historical behavior dataset is decomposed by a preset matrix factorization method to obtain a result matrix of preset data dimensions and a one-dimensional coefficient matrix. The result matrix and the coefficient matrix are non-negative matrices. The preset data dimensions are consistent with the number of data sources from which the historical behavior dataset originates. The preset matrix factorization method is a non-negative matrix factorization algorithm. The decomposition matrix includes the result matrix. The step of determining the variable fusion coefficients corresponding to each variable from each of the data sources in the historical behavior dataset based on the historical behavior dataset and the decomposition matrix includes: The historical behavior dataset is used as the input value of the regression model, and the result matrix is ​​used as the input label corresponding to the input value of the regression model. Based on the input values ​​of the regression model and the input labels, the variable fusion coefficients corresponding to each variable from each of the data sources in the historical behavior dataset are obtained by fitting a multiple linear regression model.

2. The credit risk prediction method of claim 1, wherein, The step of integrating the variables into historical behavior fusion data based on the fusion coefficients of each variable includes: Determine the product between each variable and its corresponding fusion coefficient; The products mentioned above are integrated into historical behavior fusion data.

3. The credit risk prediction method of claim 1, wherein, Before the step of performing credit risk prediction on the user based on the historical behavior fusion data and risk prediction model to obtain the risk prediction result, the method further includes: Obtain the training behavior dataset and the corresponding real labels of the training behavior dataset, wherein the training behavior dataset comes from at least two data sources; The training behavior dataset is integrated into training behavior fusion data; Based on the training behavior fusion data and the real labels, the risk prediction model to be trained is iteratively optimized to obtain the risk prediction model.

4. A credit risk prediction device, characterized in that, The credit risk prediction device includes: The acquisition module is used to acquire the user's historical behavior dataset, wherein the historical behavior dataset comes from at least two data sources; The integration module is used to integrate the historical behavior dataset into historical behavior fusion data; The prediction module is used to predict the credit risk of the user based on the historical behavior fusion data and the risk prediction model, and obtain the risk prediction result. The integration module is further configured to: determine the variable fusion coefficients corresponding to each variable from each of the data sources in the historical behavior dataset; and integrate the variables into historical behavior fusion data based on the variable fusion coefficients. The integration module is further configured to: decompose the historical behavior dataset to obtain a decomposition matrix; and determine the variable fusion coefficients corresponding to each variable in the historical behavior dataset originating from each of the data sources, based on the historical behavior dataset and the decomposition matrix. The decomposition matrix includes a result matrix and a coefficient matrix. The integration module is further configured to: decompose the historical behavior dataset using a preset matrix decomposition method to obtain a result matrix of preset data dimensions and a one-dimensional coefficient matrix, wherein the result matrix and the coefficient matrix are non-negative matrices, the preset data dimensions are consistent with the number of data sources from which the historical behavior dataset originates, and the preset matrix decomposition method is a non-negative matrix decomposition algorithm. The integration module is further configured to: use the historical behavior dataset as the input value of the regression model, and use the result matrix as the input label corresponding to the input value of the regression model; and, based on the input value of the regression model and the input label, obtain the variable fusion coefficients corresponding to each variable from each of the data sources in the historical behavior dataset by fitting a multiple linear regression model.

5. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the steps of the credit risk prediction method according to any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program for implementing a credit risk prediction method, which is executed by a processor to implement the steps of the credit risk prediction method as described in any one of claims 1 to 3.

Images