Model training method, risk prediction method and device

Abstract

The present application provides a model training method, a risk prediction method and a device, which are applied to the field of financial technology. The present application determines multiple sample data sets based on the overlap of customer data of sample customers, and then trains micro-risk models based on each sample data set, thereby making full use of the available data of various dimensions and overcoming the problem of missing block data or dimensional data in the model training process; in addition, the present application improves data utilization, and uses as much data as possible to evaluate the customer's default risk, which can improve the accuracy of the evaluation.

Description

Technical Field

[0001] The present application relates to the field of financial technology. Specifically, the present application relates to a model training method, a risk prediction method and a device. Background Art

[0002] Under the new financial form, the Internet and big data have promoted the upgrading of consumption structure, given rise to technological innovation and financial innovation, promoted the innovation of credit products, and improved the financial product system. Both online products and offline businesses have achieved rapid development, but at the same time have also given rise to new credit management issues.

[0003] On the one hand, different scenarios or segmented products of banks often provide different dimensions of corporate or personal information, but the sources of different data are closely related to the regions, customer groups and periods applicable to specific scenarios, resulting in the problem of missing data when building risk models. On the other hand, traditional financial institutions often develop risk assessment models for different segmented products separately, or eliminate low-coverage features to build a unified risk assessment model.

[0004] However, the problem with the former is that for products with smaller business volumes, the number of training samples is low, and the model built based on data-driven construction is not stable; while the latter's unified traditional scoring card model built for multiple business scenarios requires that the input features can cover most customer groups, and cannot fully and effectively utilize customer data unique to different business scenarios. Summary of the invention

[0005] This application provides a model training method, a risk prediction method and a device for making full use of customer data, improving data utilization and improving the accuracy of risk prediction. The technical solutions adopted in this application are as follows:

[0006] In a first aspect, a model training method is provided, comprising:

[0007] Obtain customer data and default information for multiple sample customers. The dimensions of customer data for different sample customers are not exactly the same;

[0008] Determine a plurality of sample data sets based on the overlap of the customer data of the sample customers;

[0009] Micro risk models are trained based on each of the sample data sets to obtain multiple micro risk models.

[0010] Optionally, the determining a plurality of sample data sets based on the overlap of the customer data of the sample customers includes:

[0011] Determine the dimensions of data included in the customer data based on the customer data of all sample customers;

[0012] A sample data set is determined based on any at least two data dimensions.

[0013] Optionally, determining a sample data set based on any at least two data dimensions includes:

[0014] Traversing the customer data of the sample customer;

[0015] When the customer data of the sample customer includes data of the at least two data dimensions, the customer data of the sample customer and the corresponding default information are added to the sample data set corresponding to the at least two data dimensions.

[0016] Optionally, the training of micro risk models based on each of the sample data sets to obtain multiple micro risk models includes:

[0017] Determining whether the customer data in each of the sample data sets meets a predetermined quantity threshold condition;

[0018] If a certain sample data set meets a predetermined quantity threshold condition, a micro risk model is trained based on the sample data in the certain sample data set.

[0019] Optionally, the method further comprises:

[0020] Based on the customer data and default information of the multiple sample customers, the weights of each micro risk model are determined through training.

[0021] Second, a risk prediction method is provided, comprising:

[0022] Obtain customer data of target customers;

[0023] Determining at least one target micro risk model matching the target customer from a plurality of micro risk models trained according to the first aspect based on the customer data of the target customer;

[0024] A target micro default risk of the target customer is determined based on the determined at least one target micro risk model and the customer data of the target customer.

[0025] Optionally, determining at least one target micro risk model matching the target customer from a plurality of micro risk models trained according to the first aspect based on the customer data of the target customer comprises:

[0026] Determining data dimensions included in the target customer based on the customer data of the target customer;

[0027] A target micro risk model is determined based on the data dimensions included in the target customer and the data dimensions of the training sample data corresponding to the pre-trained micro risk model.

[0028] Optionally, the method further comprises:

[0029] When the target micro risk model includes multiple ones, the target micro default risk is determined based on the risk prediction results of each target micro risk model and the corresponding weight of each target micro risk model.

[0030] Optionally, the method further comprises:

[0031] Normalize the weights of each target micro-risk model to obtain the normalized weights of each target micro-risk model;

[0032] The target micro default risk is determined based on the risk prediction results of each target micro risk model and the corresponding weight of each target micro risk model, including:

[0033] The target micro default risk is determined based on the risk prediction results of each target micro risk model and the normalized weight of each target micro risk model.

[0034] Optionally, the normalizing the weights of each target micro risk model to obtain the normalized weights of each target micro risk model includes:

[0035] Obtain the weight value of each target micro-risk model based on the formula Determine the normalized weight value of each target micro-model, where X k represents the weight of the target micro-model, K represents the number of target micro-models, ω k Represents the normalized weight value of each target micro-risk model.

[0036] Optionally, the method is characterized by comprising:

[0037] Determine the target macro default risk based on the pre-trained macro risk model;

[0038] The target default risk of the target customer is determined based on the determined target macro default risk and the target micro default risk.

[0039] Optionally, determining a target macro default risk based on a pre-trained macro risk model includes:

[0040] Obtain current macro-economic indicator data;

[0041] Based on the current macro- and meso-economic indicator data obtained, the target macro default risk is determined through a pre-trained macro risk model.

[0042] In a third aspect, a model training device is provided, characterized in that it includes:

[0043] The first acquisition module is used to acquire customer data and default information of multiple sample customers, where the dimensions of customer data of different sample customers are not completely the same;

[0044] A first determining module, configured to determine a plurality of sample data sets based on the coincidence of the customer data of the sample customers acquired by the first acquiring module;

[0045] The first training module is used to train micro risk models based on each of the sample data sets determined by the first determination module to obtain multiple micro risk models.

[0046] Optionally, the first determining module includes:

[0047] A first determining unit, configured to determine the dimension of data included in the customer data based on the customer data of all sample customers;

[0048] The second determining unit determines a sample data set based on any at least two data dimensions.

[0049] Optionally, the second determining unit is specifically configured to traverse the customer data of the sample customer;

[0050] And when the customer data of the sample customer includes data of the at least two data dimensions, the customer data of the sample customer and the corresponding default information are added to the sample data set corresponding to the at least two data dimensions.

[0051] Optionally, the device further comprises:

[0052] A judgment module, used to judge whether the customer data in each of the sample data sets meets a predetermined quantity threshold condition;

[0053] The training module is specifically used to train a micro risk model based on sample data in a certain sample data set if the certain sample data set meets a predetermined quantity threshold condition.

[0054] Optionally, the device further comprises:

[0055] The second training module is specifically used to train and determine the weights of each micro risk model based on the customer data and default information of the multiple sample customers.

[0056] In a fourth aspect, a risk prediction device is provided, the device comprising:

[0057] The second acquisition module is used to acquire customer data of target customers;

[0058] A second determination module, configured to determine at least one target micro risk model matching the target customer from a plurality of micro risk models trained according to the first aspect based on the customer data of the target customer;

[0059] The third determination module is used to determine the target micro default risk of the target customer based on at least one target micro risk model determined by the second determination module and the customer data of the target customer.

[0060] Optionally, the second determining module includes:

[0061] A third determining unit, configured to determine the data dimensions included in the target customer based on the customer data of the target customer;

[0062] The fourth determination unit is used to determine the target micro risk model based on the data dimensions included in the target customer and the data dimensions of the training sample data corresponding to the pre-trained micro risk model.

[0063] Optionally, the device further comprises:

[0064] The fourth determination module is used to determine the target micro default risk based on the risk prediction results of each target micro risk model and the corresponding weight of each target micro risk model when the target micro risk model includes multiple ones.

[0065] Optionally, the device further comprises:

[0066] A normalization module is used to normalize the weights of each target micro-risk model to obtain the normalized weights of each target micro-risk model;

[0067] The fourth determination module is specifically used to determine the target micro default risk based on the risk prediction results of each target micro risk model and the normalized weight of each target micro risk model.

[0068] Optionally, the normalization module is specifically used to obtain the weight value of each target micro-risk model based on the formula Determine the normalized weight value of each target micro-model, where X k represents the weight of the target micro-model, K represents the number of target micro-models, ω k Represents the normalized weight value of each target micro-risk model.

[0069] Optionally, the device further comprises:

[0070] A fifth determination module, for determining a target macro default risk based on a pre-trained macro risk model;

[0071] The sixth determination module is used to determine the target default risk of the target customer based on the determined target macro default risk and the target micro default risk.

[0072] Optionally, the device further comprises:

[0073] The third acquisition module is used to obtain the current macro- and meso-economic indicator data;

[0074] The seventh determination module is used to determine the target macro default risk through a pre-trained macro risk model based on the current macro- and meso-economic indicator data obtained.

[0075] In a fifth aspect, an electronic device is provided, the electronic device comprising:

[0076] one or more processors;

[0077] Memory;

[0078] One or more applications, wherein the one or more applications are stored in a memory and configured to be executed by one or more processors, and the one or more programs are configured to: execute the method shown in the first aspect and / or the second aspect.

[0079] In a sixth aspect, a computer-readable storage medium is provided, which is used to store computer instructions. When the computer-readable storage medium is run on a computer, the computer can execute the method shown in the first aspect and / or the second aspect.

[0080] The present application provides a model training method, a risk prediction method and a device. The present application obtains the customer data and default information of multiple sample customers, and the dimensions of the customer data of different sample customers are not exactly the same; multiple sample data sets are determined based on the overlap of the customer data of the sample customers; micro-risk models are trained based on each of the sample data sets to obtain multiple micro-risk models. That is, the present application determines multiple sample data sets based on the overlap of the customer data of the sample customers, and trains micro-risk models based on each of the sample data sets, so that the available data of each dimension can be fully utilized to overcome the problem of missing block data or dimensional data in the model training process. In addition, the present application improves data utilization, and uses as much data as possible available to the customer to evaluate the customer's default risk, thereby improving the accuracy of the evaluation.

[0081] Additional aspects and advantages of the present application will be partially given in the following description, which will become apparent from the following description, or will be understood through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0083] Figure 1 A flowchart of a model training method according to an embodiment of the present application;

[0084] Figure 2 A schematic diagram of a risk prediction method according to an embodiment of the present application;

[0085] Figure 3 A schematic diagram of the structure of a model training device according to an embodiment of the present application;

[0086] Figure 4 A schematic diagram of the structure of a risk prediction device according to an embodiment of the present application;

[0087] Figure 5 A schematic diagram of the structure of an electronic device according to an embodiment of the present application;

[0088] Figure 6 This is an example diagram of a debt sample according to an embodiment of the present application. DETAILED DESCRIPTION

[0089] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.

[0090] It will be understood by those skilled in the art that, unless otherwise stated, the singular forms "a", "an" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0091] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0092] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0093] Embodiment 1

[0094] The present application embodiment provides a model training method, such as Figure 1 As shown, the method may include the following steps:

[0095] Step S101, obtaining customer data and default information of multiple sample customers, where the dimensions of customer data of different sample customers are not completely the same.

[0096] Specifically, the sample customer can be an individual customer or a corporate customer, and the corresponding customer data can be the customer's basic information, such as name, company name, transaction information, business information, historical settlement data of the company, corporate credit data, corporate tax data, settlement data of the actual controller of the company, credit data of the actual controller of the company, consumption data of the actual controller of the company, etc. Different data dimensions can be different data types or different data sources, such as consumption data or business data of users or companies obtained through a certain platform. Among them, the default information refers to whether the customer has defaulted.

[0097] Among them, the customer data of the obtained sample customers can also be cleaned based on the historical sample data distribution, the range of outliers can be determined, and missing values, outliers and standardization can be performed.

[0098] Step S102: determining a plurality of sample data sets based on the overlap of the customer data of the sample customers.

[0099] For example, sample customer A includes data in four dimensions of ABCD, sample customer B includes data in four dimensions of ABDE, and sample customer C includes data in four dimensions of ABCE. A, B, and C each contain data in two dimensions of AB, and a sample data set can be determined based on AB; A and B each contain data in three dimensions of ABD, and a sample data set can be determined based on ABD; B and C each contain data in three dimensions of ABE, and a sample data set can be determined based on ABE; based on the above method, the relevant customer data can be added to the corresponding sample data set according to the overlap between the customer data of other sample customers, thereby obtaining multiple sample data sets.

[0100] Specifically, the data of each dimension or multiple dimensions can be regarded as a data block, and multiple sample data sets can be determined based on the overlap of the data blocks. Specifically, the data of multiple related dimensions can be regarded as a data block based on the correlation of the data dimensions. Specifically, for each data module, features of different dimensions derived from this type of data can be constructed separately; for each data module, the predictive ability of the variable can be tested for the historical sample data covered by each feature, and the variables with poor predictive ability can be deleted from the full variable list, and a new variable set can be constructed as the variable set used by the final model. Specifically, the Lasso algorithm can be used to automatically screen features with certain predictive abilities.

[0101] For example, Figure 6 As shown, statistics are collected for each debt sample (i.e., customer sample) to determine that a total of three data blocks, GS1, GS2, and GS3, are included. Debt sample 1 (i.e., sample customer 1) includes data blocks GS1 and GS2, debt sample 2 (i.e., sample customer 2) includes data blocks GS1, GS2, and GS3, and debt sample L includes data blocks GS2 and GS3. Based on the data overlap between customer data, that is, the overlap between databases, a sample data set is determined. For sample customer 1 and sample customer 2, GS1 and GS2 are data blocks included by both, and a sample data set can be determined based on databases GS1 and GS2; for sample customer 2 and sample customer L, GS2 and GS3 are data blocks included by both, and a sample data set can be determined based on databases GS1 and GS2.

[0102] Step S103: training a micro risk model based on each of the sample data sets to obtain a plurality of micro risk models.

[0103] Specifically, the micro risk model can be trained based on each of the sample data sets using different algorithms such as logistic regression and random forest. Preferably, the micro risk model can be trained using the LightGBM algorithm.

[0104] The embodiment of the present application provides a possible implementation method. Specifically, the determining of multiple sample data sets based on the overlap of customer data of the sample customers includes:

[0105] Determine the dimensions of data included in the customer data based on the customer data of all sample customers;

[0106] A sample data set is determined based on any at least two data dimensions.

[0107] Exemplarily, customer A includes data of two dimensions AB, customer B includes data of two dimensions AC, and customer C includes data of three dimensions ABC, then the dimensions of the data contained in the customer data are determined to be the three dimensions ABC; among which four sample number sets AB / AC / BC / ABC can be determined.

[0108] The embodiment of the present application provides a possible implementation method. Specifically, a sample data set is determined based on any at least two data dimensions, including:

[0109] Traversing the customer data of the sample customer;

[0110] When the customer data of the sample customer includes data of the at least two data dimensions, the customer data of the sample customer and the corresponding default information are added to the sample data set corresponding to the at least two data dimensions.

[0111] The embodiment of the present application provides a possible implementation method. Specifically, the micro risk models are trained based on each of the sample data sets to obtain multiple micro risk models, which includes:

[0112] Determining whether the customer data in each of the sample data sets meets a predetermined quantity threshold condition;

[0113] If a certain sample data set meets a predetermined quantity threshold condition, a micro risk model is trained based on the sample data in the certain sample data set.

[0114] Specifically, if the customer data in the sample data set meets a predetermined threshold condition, such as greater than 50, the micro risk model is trained based on the sample data in the sample data set. Specifically, the number of defaults of customers in the sample data set for training the micro risk model may be limited to be greater than a predetermined threshold.

[0115] Optionally, the method further comprises:

[0116] Based on the customer data and default information of the plurality of sample customers, the weights of each micro risk model are determined by training. Specifically, the weights of each micro risk model can be determined by training with a corresponding deep learning algorithm. Specifically, the weights of each micro risk model can be optimized by training with the goal of maximizing the model prediction performance under historical samples.

[0117] The present application provides a model training method, which obtains customer data and default information of multiple sample customers, where the dimensions of customer data of different sample customers are not exactly the same; determines multiple sample data sets based on the overlap of the customer data of the sample customers; and trains micro-risk models based on each of the sample data sets to obtain multiple micro-risk models. That is, the present application determines multiple sample data sets based on the overlap of the customer data of the sample customers, and trains micro-risk models based on each of the sample data sets, thereby making full use of the available data of each dimension and overcoming the problem of missing block data or dimensional data in the model training process. In addition, the present application improves data utilization, uses all the data available to the customer as much as possible to evaluate the customer's default risk, and improves the accuracy of the evaluation.

[0118] Embodiment 2

[0119] like Figure 2 As shown, the embodiment of the present application provides a risk prediction method, the method comprising:

[0120] Step S201, obtaining customer data of target customers;

[0121] Step S202, determining at least one target micro risk model matching the target customer from a plurality of micro risk models trained according to the first aspect based on the customer data of the target customer;

[0122] Specifically, determining at least one target micro risk model matching the target customer from a plurality of micro risk models trained according to the first aspect based on the customer data of the target customer comprises:

[0123] Determining data dimensions included in the target customer based on the customer data of the target customer;

[0124] A target micro risk model is determined based on the data dimensions included in the target customer and the data dimensions of the training sample data corresponding to the pre-trained micro risk model.

[0125] Exemplarily, the data of the target customer contains two dimensions, AB. The data dimension for training micro-risk model 1 is ABC, the data dimension for training micro-risk model 2 is AC, the data dimension for training micro-risk model 3 is AB, and the data dimension for training micro-risk model 4 is BC. Then, the micro-risk model that matches the target customer is determined to be micro-risk model 3.

[0126] For example, for example, the data of the target customer contains three dimensions of ABC, the data dimension for training micro-risk model 1 is ABC, the data dimension for training micro-risk model 2 is AC, the data dimension for training micro-risk model 3 is AB, the data dimension for training micro-risk model 4 is BC, and the data dimension for training micro-risk model 5 is ABCDE, then the micro-risk models matching the target customer are determined to be micro-risk model 1, micro-risk model 2, micro-risk model 3, and micro-risk model 4.

[0127] Step S203: determining the target micro default risk of the target customer based on the determined at least one target micro risk model and the customer data of the target customer.

[0128] Specifically, when the target micro-risk model includes multiple ones, the target micro-default risk is determined based on the risk prediction results of each target micro-risk model and the corresponding weight of each target micro-risk model; specifically, the weight can be manually determined through prior knowledge, or obtained through training, such as training through deep learning methods.

[0129] The embodiment of the present application provides a possible implementation method. Further, if each target micro-risk model is obtained through training, the method further includes:

[0130] Normalize the weights of each target micro-risk model to obtain the normalized weights of each target micro-risk model;

[0131] The target micro default risk is determined based on the risk prediction results of each target micro risk model and the corresponding weight of each target micro risk model, including:

[0132] The target micro default risk is determined based on the risk prediction results of each target micro risk model and the normalized weight of each target micro risk model.

[0133] Optionally, the normalizing the weights of each target micro risk model to obtain the normalized weights of each target micro risk model includes:

[0134] Obtain the weight value of each target micro-risk model based on the formula Determine the normalized weight value of each target micro-model, where X k represents the weight of the target micro-model, K represents the number of target micro-models, ω k Represents the normalized weight value of each target micro-risk model.

[0135] The embodiment of the present application provides a possible implementation manner. Further, the method also includes:

[0136] Determine the target macro default risk based on the pre-trained macro risk model;

[0137] The target default risk of the target customer is determined based on the determined target macro default risk and the target micro default risk.

[0138] For the embodiments of the present application, by integrating the micro risk model and the macro risk model to determine the default risk of the target customer, the accuracy and stability of the prediction can be improved.

[0139] Optionally, determining a target macro default risk based on a pre-trained macro risk model includes:

[0140] Obtain current macro- and meso-economic indicator data; specifically, macro- and meso-economic indicators may be GDP indicators, CPI indicators of various provinces, PPI year-on-year, industrial added value year-on-year, electricity consumption year-on-year, RMB to USD exchange rate, etc. Specifically, macro- and meso-economic indicators may be screened based on their predictability.

[0141] Based on the current macro- and meso-economic indicator data obtained, the target macro default risk is determined through a pre-trained macro risk model.

[0142] For the embodiment of the present application, a model of the corresponding data dimension is determined based on the customer data of the target customer, and the customer's default risk is evaluated using all the data available to the customer as much as possible, thereby improving data utilization and increasing the accuracy of the evaluation.

[0143] Embodiment 3

[0144] The present application embodiment provides a model training device, the device 30 comprising:

[0145] The first acquisition module 301 is used to acquire customer data and default information of multiple sample customers, where the dimensions of customer data of different sample customers are not completely the same;

[0146] A first determining module 302, configured to determine a plurality of sample data sets based on the coincidence of the customer data of the sample customers acquired by the first acquiring module;

[0147] The first training module 303 is used to train micro risk models based on each of the sample data sets determined by the first determination module to obtain multiple micro risk models.

[0148] Optionally, the first determining module includes:

[0149] A first determining unit, configured to determine the dimension of data included in the customer data based on the customer data of all sample customers;

[0150] The second determining unit determines a sample data set based on any at least two data dimensions.

[0151] Optionally, the second determining unit is specifically configured to traverse the customer data of the sample customer;

[0152] And when the customer data of the sample customer includes data of the at least two data dimensions, the customer data of the sample customer and the corresponding default information are added to the sample data set corresponding to the at least two data dimensions.

[0153] Optionally, the device further comprises:

[0154] A judgment module, used to judge whether the customer data in each of the sample data sets meets a predetermined quantity threshold condition;

[0155] The training module is specifically used to train a micro risk model based on sample data in a certain sample data set if the certain sample data set meets a predetermined quantity threshold condition.

[0156] Optionally, the device further comprises:

[0157] The second training module is specifically used to train and determine the weights of each micro risk model based on the customer data and default information of the multiple sample customers.

[0158] The device of the embodiment of the present application can execute the method of the first embodiment. The beneficial effects of the embodiment of the present application are the same as those of the first embodiment and will not be described in detail here.

[0159] Embodiment 4

[0160] The embodiment of the present application provides a risk prediction device, the device 40 comprising:

[0161] The second acquisition module 401 is used to acquire customer data of target customers;

[0162] A second determination module 402 is used to determine at least one target micro risk model matching the target customer from a plurality of micro risk models trained according to the first aspect based on the customer data of the target customer;

[0163] The third determination module 403 is used to determine the target micro default risk of the target customer based on at least one target micro risk model determined by the second determination module and the customer data of the target customer.

[0164] Optionally, the second determining module includes:

[0165] A third determining unit, configured to determine the data dimensions included in the target customer based on the customer data of the target customer;

[0166] The fourth determination unit is used to determine the target micro risk model based on the data dimensions included in the target customer and the data dimensions of the training sample data corresponding to the pre-trained micro risk model.

[0167] Optionally, the device further comprises:

[0168] The fourth determination module is used to determine the target micro default risk based on the risk prediction results of each target micro risk model and the corresponding weight of each target micro risk model when the target micro risk model includes multiple ones.

[0169] Optionally, the device further comprises:

[0170] A normalization module is used to normalize the weights of each target micro-risk model to obtain the normalized weights of each target micro-risk model;

[0171] The fourth determination module is specifically used to determine the target micro default risk based on the risk prediction results of each target micro risk model and the normalized weight of each target micro risk model.

[0172] Optionally, the normalization module is specifically used to obtain the weight value of each target micro-risk model based on the formula Determine the normalized weight value of each target micro-model, where X k represents the weight of the target micro-model, K represents the number of target micro-models, ω k Represents the normalized weight value of each target micro-risk model.

[0173] Optionally, the device further comprises:

[0174] A fifth determination module, for determining a target macro default risk based on a pre-trained macro risk model;

[0175] The sixth determination module is used to determine the target default risk of the target customer based on the determined target macro default risk and the target micro default risk.

[0176] Optionally, the device further comprises:

[0177] The third acquisition module is used to obtain the current macro- and meso-economic indicator data;

[0178] The seventh determination module is used to determine the target macro default risk through a pre-trained macro risk model based on the current macro- and meso-economic indicator data obtained.

[0179] The beneficial effects of the embodiment of the present application are the same as those of the second embodiment and will not be described in detail here.

[0180] Embodiment 5

[0181] The present application embodiment provides an electronic device, such as Figure 5 As shown, Figure 5 The electronic device 50 shown includes: a processor 501 and a memory 505. The processor 501 and the memory 503 are connected, such as through a bus 502. Further, the electronic device 50 may also include a transceiver 503. It should be noted that in actual applications, the transceiver 504 is not limited to one, and the structure of the electronic device 50 does not constitute a limitation on the embodiments of the present application. The processor 501 is used in the embodiments of the present application to implement Figure 3 The transceiver 504 includes a receiver and a transmitter.

[0182] Processor 501 may be a CPU, a general processor, a DSP, an ASIC, an FPGA or other programmable logic device, a transistor logic device, a hardware component or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 501 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0183] The bus 502 may include a path to transmit information between the above components. The bus 502 may be a PCI bus or an EISA bus, etc. The bus 502 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0184] The memory 505 can be a ROM or other type of static storage device that can store static information and instructions, a RAM or other type of dynamic storage device that can store information and instructions, or an EEPROM, a CD-ROM or other optical disk storage, an optical disk storage (including a compressed optical disk, a laser disk, an optical disk, a digital versatile disk, a Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to this.

[0185] The memory 505 is used to store the application code for executing the solution of the present application, and the execution is controlled by the processor 501. The processor 501 is used to execute the application code stored in the memory 505 to implement Figure 3 or Figure 4 The illustrated embodiments provide for the functionality of the device.

[0186] The embodiment of the present application provides an electronic device. The present application determines multiple sample data sets based on the coincidence of the customer data of the sample customers, and trains micro-risk models based on each sample data set, so as to make full use of the available dimensional data and overcome the problem of missing block data or dimensional data in the model training process. In addition, the present application improves data utilization, and uses all the data available to the customer to evaluate the customer's default risk as much as possible, thereby improving the accuracy of the evaluation.

[0187] The present application embodiment provides an electronic device applicable to the above method embodiment, which will not be described in detail here.

[0188] Embodiment 6

[0189] An embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the method shown in the above embodiment is implemented.

[0190] The embodiment of the present application provides a computer-readable storage medium. The present application determines multiple sample data sets based on the coincidence of the customer data of the sample customers, and trains micro-risk models based on each sample data set, so as to make full use of the available dimensional data and overcome the problem of missing block data or dimensional data in the model training process. In addition, the present application improves data utilization, and uses all the data available to the customer to evaluate the customer's default risk as much as possible, thereby improving the accuracy of the evaluation.

[0191] The present application embodiment provides a computer-readable storage medium applicable to the above method embodiment, which will not be described in detail here.

[0192] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0193] The above are only some implementation methods of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A risk prediction method, characterized in that: include: Obtain customer data of target customers; Determining at least one target micro risk model matching the target customer from a plurality of trained micro risk models based on the customer data of the target customer; Determining a target micro default risk of the target customer based on the determined at least one target micro risk model and the customer data of the target customer; When the target micro risk model includes multiple ones, the target micro default risk is determined based on the risk prediction results of each target micro risk model and the corresponding weight of each target micro risk model; The training method of the micro risk model includes: Obtain customer data and default information for multiple sample customers. The dimensions of customer data for different sample customers are not exactly the same; Determining multiple sample data sets based on the overlap of the customer data of the sample customers, including determining the dimensions of data included in the customer data based on the customer data of all sample customers; and determining a sample data set based on any at least two data dimensions; Based on each of the sample data sets, a micro risk model is trained respectively to obtain a plurality of micro risk models; Based on the customer data and default information of multiple sample customers, the weights of each micro risk model are determined through training.

2. The method according to claim 1, characterized in that: A sample data set is determined based on any at least two data dimensions, including: Traversing the customer data of the sample customer; When the customer data of the sample customer includes data of the at least two data dimensions, the customer data of the sample customer and the corresponding default information are added to the sample data set corresponding to the at least two data dimensions.

3. The method according to claim 1, characterized in that The micro risk models are trained based on each of the sample data sets to obtain multiple micro risk models, which includes: Determining whether the customer data in each of the sample data sets meets a predetermined quantity threshold condition; If a certain sample data set meets a predetermined quantity threshold condition, a micro risk model is trained based on the sample data in the certain sample data set.

4. The method according to claim 1, characterized in that: Determining at least one target micro risk model matching the target customer from a plurality of trained micro risk models based on the customer data of the target customer, comprising: Determining data dimensions included in the target customer based on the customer data of the target customer; A target micro risk model is determined based on the data dimensions included in the target customer and the data dimensions of the training sample data corresponding to the pre-trained micro risk model.

5. The method according to claim 1, characterized in that The method further includes: Normalize the weights of each target micro-risk model to obtain the normalized weights of each target micro-risk model; The target micro default risk is determined based on the risk prediction results of each target micro risk model and the corresponding weight of each target micro risk model, including: The target micro default risk is determined based on the risk prediction results of each target micro risk model and the normalized weight of each target micro risk model.

6. The method according to claim 5, characterized in that The weights of each target micro-risk model are normalized to obtain the normalized weights of each target micro-risk model, including: Obtain the weight value of each target micro-risk model based on the formula Determine the normalized weight value of each target micro-model, where X k represents the weight of the target micro-model, K represents the number of target micro-models, ω k Represents the normalized weight value of each target micro-risk model.

7. The method according to claim 1, characterized in that include: Determine the target macro default risk based on the pre-trained macro risk model; The target default risk of the target customer is determined based on the determined target macro default risk and the target micro default risk.

8. The method according to claim 7, characterized in that Determine the target macro default risk based on the pre-trained macro risk model, including: Obtain current macro-economic indicator data; Based on the current macro- and meso-economic indicator data obtained, the target macro default risk is determined through a pre-trained macro risk model.

9. A risk prediction device, characterized in that: include: The second acquisition module is used to acquire customer data of target customers; A second determination module is used to determine at least one target micro risk model matching the target customer from a plurality of trained micro risk models based on the customer data of the target customer; A third determination module, configured to determine a target micro default risk of the target customer based on at least one target micro risk model determined by the second determination module and the customer data of the target customer; A fourth determination module, for determining the target micro default risk based on the risk prediction results of each target micro risk model and the corresponding weight of each target micro risk model when the target micro risk model includes multiple ones; The training method of the micro risk model includes: Obtain customer data and default information for multiple sample customers. The dimensions of customer data for different sample customers are not exactly the same; Determining multiple sample data sets based on the overlap of the customer data of the sample customers, including determining the dimensions of data included in the customer data based on the customer data of all sample customers; and determining a sample data set based on any at least two data dimensions; Based on each of the sample data sets, a micro risk model is trained respectively to obtain a plurality of micro risk models; Based on the customer data and default information of multiple sample customers, the weights of each micro risk model are determined through training.

10. An electronic device, characterized in that: include: one or more processors; Memory; One or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to: execute the method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that: The computer storage medium is used to store computer instructions, and when the computer storage medium is run on a computer, the computer can execute the method according to any one of claims 1 to 8.

Images