Parameter determination method and device, program product and storage medium

Abstract

The embodiment of the invention provides a parameter determination method and device, a program product and a storage medium, and the method comprises the steps: obtaining a plurality of risk index data of a target object, the risk index data being used for describing the credit risk of the target object; inputting the multiple pieces of risk index data into a target classification model to obtain a target risk factor, the target classification model being obtained by training an initial classification model by using multiple pieces of training risk index data of multiple training objects, and the initial classification model comprising a classification sub-model and an adjustment sub-model; and determining a target parameter of the target object by using the target risk factor. According to the method and the device, the problem that the credit granting related parameters cannot be accurately determined in related technologies is solved, and the effect of accurately determining the credit granting related parameters is realized.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, and more specifically, to a method and device for determining a parameter, a program product, and a storage medium. Background Art

[0002] In the financial industry, credit-related parameters (for example, the credit limit of customers in the banking industry) are important parameters for assessing customer credit status and controlling risks, and the accuracy and efficiency of their calculation are of great significance. Related technologies usually use machine learning models, such as linear regression, to calculate credit-related parameters. However, traditional machine learning models often only focus on local features or single samples in the calculation of credit-related parameters, and cannot consider the distribution of credit-related parameters from a global perspective, resulting in the problem of being unable to accurately determine credit-related parameters. Summary of the invention

[0003] The embodiments of the present application provide a method and device for determining parameters, a program product, and a storage medium to solve the problem that the parameters related to credit cannot be accurately determined in the related art.

[0004] According to an embodiment of the present application, a parameter determination method is provided, comprising: obtaining multiple risk indicator data of a target object, wherein the risk indicator data is used to describe the credit risk of the target object; inputting the multiple risk indicator data into a target classification model to obtain a target risk factor, wherein the target classification model is obtained by training an initial classification model using multiple training risk indicator data of multiple training objects, the initial classification model comprises a classification sub-model and an adjustment sub-model, the classification sub-model is used to generate a multi-dimensional three-dimensional grid based on the multiple training risk indicator data, the adjustment sub-model is used to adjust the initial training risk indicator coefficients of the multiple training risk indicator data in the multi-dimensional three-dimensional grid under the condition of satisfying constraints, so that the bad debt rate of the multiple training objects is less than a first threshold value, and outputting the risk factor corresponding to each training object based on the adjusted training risk indicator coefficient; and determining the target parameters of the target object using the target risk factor.

[0005] In an exemplary embodiment, obtaining multiple risk indicator data of a target object includes: obtaining identification information of the target object, wherein the identification information is information for identifying the target object; obtaining multiple original risk indicator data of the target object from a target platform based on the identification information, wherein the original risk indicator data is used to describe the credit risk of the target object; performing target processing on the multiple original risk indicator data respectively to obtain the multiple risk indicator data, wherein the target processing is used to adjust the contribution of the risk indicator data to the target classification model.

[0006] In an exemplary embodiment, before inputting the plurality of risk indicator data into the target classification model to obtain the target risk factor, the method further includes: acquiring the original training risk indicator data of the plurality of training objects from the target platform based on the identification information of the plurality of training objects, wherein the original training risk indicator data is used to describe the credit risk of the training objects, and each of the training objects includes one or more original training risk indicator data; performing target processing on the plurality of original training risk indicator data respectively to obtain the plurality of training risk indicator data, wherein the target processing is used to adjust the contribution of the original training risk indicator data to the classification sub-model; extracting the plurality of dimensions of the plurality of training risk indicator data based on the attribute information of the plurality of training risk indicator data, wherein each dimension includes one or more training risk indicator data; generating the multi-dimensional three-dimensional grid based on the plurality of dimensions and the plurality of training risk indicator data, wherein one training object corresponds to one grid in the multi-dimensional three-dimensional grid, and one grid is composed of the plurality of training risk indicators of different dimensions.

[0007] In an exemplary embodiment, before inputting the above-mentioned multiple risk indicator data into the target classification model to obtain the target risk factor, the above-mentioned method also includes: obtaining the first original parameter, the second original parameter and the third original parameter of each of the above-mentioned training objects from the target platform based on the identification information of the above-mentioned multiple training objects, wherein the above-mentioned first original parameter is used to represent the original loan amount of the above-mentioned training object, the above-mentioned second original parameter is used to represent the original loan expenditure balance of the above-mentioned training object, and the above-mentioned third original parameter is used to represent the original non-performing loan balance of the above-mentioned training object; based on the initial training risk indicator coefficients of the above-mentioned multiple training risk indicator data of each training object in the above-mentioned multi-dimensional three-dimensional grid, calculating the initial risk factor of each of the above-mentioned training objects, wherein one of the above-mentioned training risk indicator data corresponds to one of the above-mentioned initial training risk indicator coefficients; calculating the product of the initial risk factor of each of the above-mentioned training objects and the above-mentioned first original parameter to obtain the first target parameter of each of the above-mentioned training objects, wherein the above-mentioned first target parameter is used to represent the initial predicted loan amount; based on The above-mentioned first original parameter, the above-mentioned second original parameter and the above-mentioned first target parameter are used to calculate the above-mentioned second target parameter of each of the above-mentioned training objects, wherein the above-mentioned second target parameter is used to represent the initial predicted loan expenditure balance; based on the above-mentioned second original parameter, the above-mentioned third original parameter and the above-mentioned second target parameter, the third target parameter is used to represent the initial predicted non-performing loan balance; the ratio of the first sum value to the second sum value is calculated to obtain the above-mentioned non-performing rate, wherein the above-mentioned first sum value is the sum value of the above-mentioned third target parameters corresponding to the above-mentioned multiple training objects, and the above-mentioned second sum value is the sum value of the above-mentioned second target parameters corresponding to the above-mentioned multiple training objects; under the condition of satisfying the above-mentioned constraints, the initial training risk indicator coefficient of the above-mentioned multiple training risk indicator data of each of the above-mentioned training objects in the above-mentioned multi-dimensional three-dimensional grid is adjusted to adjust the above-mentioned initial risk factor, the above-mentioned second target parameter and the above-mentioned third target parameter so that the above-mentioned non-performing rate is less than the above-mentioned first threshold value, and the risk factor of each of the above-mentioned training objects is obtained.

[0008] In an exemplary embodiment, based on the first original parameter, the second original parameter and the first target parameter, the second target parameter of each training object is calculated, including: when the first original parameter is equal to the second original parameter, the value of the first target parameter is determined as the second target parameter; when the first original parameter is greater than the second original parameter, the minimum value between the first target parameter and the second original parameter is determined as the second target parameter.

[0009] In an exemplary embodiment, based on the second original parameter, the third original parameter and the second target parameter, the third target parameter of each training object is calculated, including: when the second original parameter is equal to the third original parameter, the value of the second target parameter is determined as the third target parameter; when the first difference between the second original parameter and the third original parameter is greater than or equal to the second target parameter, the third target parameter is determined to be zero; when the first difference is less than the second target parameter, the difference between the second target parameter and the first difference is determined as the third target parameter.

[0010] In an exemplary embodiment, the target parameters of the target object are determined using the target risk factor, including: calculating the product of the target risk factor and the target original parameter of the target object to obtain the target parameter, wherein the target original parameter is used to represent the original loan amount of the target object.

[0011] According to one embodiment of the present application, a parameter determination device includes: a first acquisition module, used to acquire multiple risk indicator data of a target object, wherein the above risk indicator data is used to describe the credit risk of the above target object; a first input module, used to input the multiple risk indicator data into a target classification model to obtain a target risk factor, wherein the above target classification model is obtained by training an initial classification model using multiple training risk indicator data of multiple training objects, and the above initial classification model includes a classification sub-model and an adjustment sub-model, and the above classification sub-model is used to generate a multi-dimensional three-dimensional grid based on the multiple training risk indicator data, and the above adjustment sub-model is used to adjust the initial training risk indicator coefficients of the multiple training risk indicator data in the multi-dimensional three-dimensional grid under the condition of satisfying constraints, so that the bad rate of the multiple training objects is less than a first threshold, and output the risk factor corresponding to each training object based on the adjusted training risk indicator coefficient; a first determination module, used to determine the target parameters of the above target object using the above target risk factor.

[0012] In an exemplary embodiment, the above-mentioned first acquisition module includes: a first acquisition sub-module, used to obtain the identification information of the above-mentioned target object, wherein the above-mentioned identification information is information used to identify the above-mentioned target object; a second acquisition sub-module, used to obtain multiple original risk indicator data of the above-mentioned target object from the target platform based on the above-mentioned identification information, wherein the above-mentioned original risk indicator data is used to describe the credit risk of the above-mentioned target object; a first processing sub-module, used to perform target processing on the multiple original risk indicator data respectively to obtain the multiple risk indicator data, wherein the above-mentioned target processing is used to adjust the contribution of the above-mentioned risk indicator data to the above-mentioned target classification model.

[0013] In an exemplary embodiment, the device further includes: a second acquisition module, used to input the plurality of risk indicator data into a target classification model, and before obtaining a target risk factor, obtain the original training risk indicator data of the plurality of training objects from a target platform based on the identification information of the plurality of training objects, wherein the original training risk indicator data is used to describe the credit risk of the training object, and each of the training objects includes one or more original training risk indicator data; a first processing module, used to perform target processing on the plurality of original training risk indicator data respectively to obtain the plurality of training risk indicator data, wherein the target processing is used to adjust the contribution of the original training risk indicator data to the classification sub-model; a first extraction module, used to extract the plurality of dimensions of the plurality of training risk indicator data based on the attribute information of the plurality of training risk indicator data, wherein each dimension includes one or more training risk indicator data; a first generation module, used to generate the multi-dimensional three-dimensional grid based on the plurality of dimensions and the plurality of training risk indicator data, wherein one training object corresponds to one grid in the multi-dimensional three-dimensional grid, and one grid is composed of a plurality of training risk indicators of different dimensions.

[0014] In an exemplary embodiment, the device further includes: a third acquisition module, which is used to input the plurality of risk indicator data into the target classification model, and before obtaining the target risk factor, obtain the first original parameter, the second original parameter and the third original parameter of each of the training objects from the target platform based on the identification information of the plurality of training objects, wherein the first original parameter is used to represent the original loan amount of the training object, the second original parameter is used to represent the original loan expenditure balance of the training object, and the third original parameter is used to represent the original non-performing loan balance of the training object; a first calculation module, which is used to calculate the initial risk factor of each of the training objects based on the initial training risk indicator coefficient of the plurality of training risk indicator data of each training object in the multi-dimensional three-dimensional grid, wherein one training risk indicator data corresponds to one initial training risk indicator coefficient; a second calculation module, which is used to calculate the product of the initial risk factor of each of the training objects and the first original parameter, and obtain the first target parameter of each of the training objects, wherein the first target parameter is used to represent the initial predicted loan amount; a third calculation module, A fourth calculation module is used to calculate the third target parameter of each training object based on the first original parameter, the second original parameter and the first target parameter, wherein the second target parameter is used to represent the initial predicted loan expenditure balance; a fifth calculation module is used to calculate the ratio of the first sum value to the second sum value to obtain the non-performing rate, wherein the first sum value is the sum value of the multiple third target parameters corresponding to the multiple training objects, and the second sum value is the sum value of the multiple second target parameters corresponding to the multiple training objects; a first adjustment module is used to adjust the initial training risk indicator coefficient of the multiple training risk indicator data of each training object in the multi-dimensional three-dimensional grid under the condition of satisfying the above constraint conditions, so as to adjust the initial risk factor, the second target parameter and the third target parameter, so as to make the non-performing rate less than the first threshold value, and obtain the risk factor of each training object.

[0015] In an exemplary embodiment, the above-mentioned third calculation module includes: a first determination submodule, used to determine the value of the above-mentioned first target parameter as the above-mentioned second target parameter when the above-mentioned first original parameter is equal to the above-mentioned second original parameter; a second determination submodule, used to determine the minimum value of the above-mentioned first target parameter and the above-mentioned second original parameter as the above-mentioned second target parameter when the above-mentioned first original parameter is greater than the above-mentioned second original parameter.

[0016] In an exemplary embodiment, the fourth calculation module includes: a third determination submodule, used to determine the value of the second target parameter as the third target parameter when the second original parameter is equal to the third original parameter; a fourth determination submodule, used to determine the third target parameter as zero when the first difference between the second original parameter and the third original parameter is greater than or equal to the second target parameter; and a fifth determination submodule, used to determine the difference between the second target parameter and the first difference as the third target parameter when the first difference is less than the second target parameter.

[0017] In an exemplary embodiment, the first determination module includes: a sixth calculation submodule, used to calculate the product of the target risk factor and the target original parameter of the target object to obtain the target parameter, wherein the target original parameter is used to represent the original loan amount of the target object.

[0018] According to another embodiment of the present application, a computer program product is provided, including a computer program, wherein the computer program is configured for a processor to execute the steps of any of the above method embodiments.

[0019] According to another embodiment of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to enable a processor to execute the steps in any of the above method embodiments.

[0020] According to another embodiment of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor is configured to execute the computer program to perform the steps in any one of the method embodiments.

[0021] Through this application, multiple risk indicator data describing the credit risk of the target object are first obtained, and then the target risk factor of the multiple risk indicator data is determined using the target classification model, and finally the target parameter of the target object is determined based on the target risk factor. Since traditional machine learning models often only focus on local features or single samples in the measurement of credit-related parameters, this application can use a multi-dimensional three-dimensional grid to focus on the training risk indicator data of multiple training objects, thereby determining the target risk factor of the target object under the condition that the bad debt rate of multiple training objects is less than the first threshold. Therefore, the problem that the relevant technology cannot accurately determine the parameters related to credit is solved, and the effect of accurately determining the parameters related to credit is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1It is a hardware structure block diagram of a mobile terminal of a method for determining parameters in an embodiment of the present application;

[0023] Figure 2 is a flow chart of a method for determining parameters according to an embodiment of the present application;

[0024] Figure 3 It is a flow chart of a method for determining a parameter in a specific embodiment of the present application;

[0025] Figure 4 It is a structural block diagram of a parameter determination device according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] The embodiments of the present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0028] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 1 is a hardware structure block diagram of a mobile terminal of a method for determining parameters in an embodiment of the present application. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0029] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as a computer program corresponding to a method for determining a parameter in an embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the mobile terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0030] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0031] In this embodiment, a method for determining a parameter is provided. Figure 2 is a flow chart of a method for determining parameters according to an embodiment of the present application, such as Figure 2 As shown, the process includes the following steps:

[0032] Step S202, obtaining a plurality of risk indicator data of the target object, wherein the risk indicator data is used to describe the credit risk of the target object;

[0033] Optionally, the target objects include, but are not limited to, enterprises or individuals in the technology industry, manufacturing industry, financial industry, consumer goods industry, healthcare industry, energy industry, telecommunications industry, and Internet industry.

[0034] Optionally, the risk indicator data includes, but is not limited to, the target object's geographical data, industry data, historical credit limit data, credit score data, debt-to-income ratio data, credit record data, and asset-liability data.

[0035] Step S204, inputting the multiple risk indicator data into the target classification model to obtain the target risk factor, wherein the target classification model is obtained by training the initial classification model using the multiple training risk indicator data of the multiple training objects, and the initial classification model includes a classification sub-model and an adjustment sub-model, the classification sub-model is used to generate a multi-dimensional three-dimensional grid based on the multiple training risk indicator data, and the adjustment sub-model is used to adjust the initial training risk indicator coefficients of the multiple training risk indicator data in the multi-dimensional three-dimensional grid under the condition of satisfying the constraint conditions, so that the defect rate of the multiple training objects is less than the first threshold value, and output the risk factor corresponding to each training object based on the adjusted training risk indicator coefficient;

[0036] Optionally, the training subjects include, but are not limited to, enterprises or individuals in the technology industry, manufacturing industry, finance industry, consumer goods industry, healthcare industry, energy industry, telecommunications industry, and Internet industry.

[0037] Optionally, the training risk indicator data includes, but is not limited to, geographical data, industry data, historical credit limit data, credit score data, debt-to-income ratio data, credit record data, and asset-liability data of the training object.

[0038] Optionally, different dimensions in the multidimensional three-dimensional grid are extracted based on different training risk indicator data. For example, the training risk indicator data of training object A includes: regional data-Beijing, industry data-manufacturing, historical credit line data-good, and the training risk indicator data of training object B includes: regional data-Tianjin, industry data-health care industry, historical credit line data-good, then the dimensions in the multidimensional three-dimensional grid include: region, industry, historical credit line. For example, the training risk indicator data of training object C includes: regional data-Beijing, industry data-manufacturing, historical credit line data-good, credit score data-excellent, debt-to-income ratio data-average, then the dimensions in the multidimensional three-dimensional grid include region, industry data, historical credit line, and debt-to-income ratio.

[0039] Optionally, different training risk indicator data are collected according to the product types of different financial products, and corresponding multi-dimensional grids are generated based on the product types. For example, for personal housing loans, the training risk indicators include but are not limited to loan-to-value ratio data, credit score data, debt-to-income ratio data, and historical credit line data, so the dimensions in the multi-dimensional grid include: loan-to-value ratio, credit score, debt-to-income ratio, and historical credit line. For example, for car loans, the training risk indicators include but are not limited to: vehicle value data, loan term data, credit score data, historical repayment record data, and debt-to-income ratio data, so the dimensions in the multi-dimensional grid include: vehicle value, loan term, credit score, historical repayment record, and debt-to-income ratio.

[0040] Optionally, the constraints include, but are not limited to: initial training risk indicator coefficient and adjustment range limit of training risk indicator coefficient, monotonicity limit, difference limit, and variation range of total historical parameters of multi-dimensional grid.

[0041] Optionally, the bad rate is the minimum value of (the total predicted bad expenditure amount / the total predicted expenditure amount).

[0042] Step S206: Determine target parameters of the target object using the target risk factor.

[0043] In this embodiment, the execution subject of the above steps can be a terminal, a server, a specific processor set in the terminal or server, or a processor or processing device set relatively independently from the terminal or server, but is not limited thereto.

[0044] Through the above steps, firstly, a plurality of risk indicator data describing the credit risk of the target object are obtained, then the target risk factor of the plurality of risk indicator data is determined by using the target classification model, and finally the target parameter of the target object is determined based on the target risk factor. Since the traditional machine learning model often only focuses on local features or a single sample in the calculation of credit-related parameters, the present application can use a multi-dimensional three-dimensional grid to focus on the training risk indicator data of multiple training objects, thereby determining the target risk factor of the target object under the condition that the bad debt rate of multiple training objects is less than the first threshold value. Therefore, the problem that the parameters related to credit cannot be accurately determined in the related technology is solved, and the effect of accurately determining the parameters related to credit is achieved.

[0045] In an exemplary embodiment, obtaining multiple risk indicator data of a target object includes: obtaining identification information of the target object, wherein the identification information is information for identifying the target object; obtaining multiple original risk indicator data of the target object from a target platform based on the identification information, wherein the original risk indicator data is used to describe the credit risk of the target object; performing target processing on the multiple original risk indicator data respectively to obtain the multiple risk indicator data, wherein the target processing is used to adjust the contribution of the risk indicator data to the target classification model.

[0046] Optionally, the target platform includes but is not limited to a database or cloud server of a financial institution. The identification information of the target object includes but is not limited to the identity information of the target object and the account information of the target object on the target platform. The original risk indicator data includes but is not limited to the target object's geographical data, industry data, historical credit limit data, credit score data, debt-to-income ratio data, credit record data, asset-liability data, enterprise scale data, enterprise personnel data, age data, educational data, gender data, and family data. Target processing is used to improve the contribution of risk indicator data to the target classification model, including but not limited to normalization processing, standardization processing, dimension removal processing, outlier processing, missing value processing, feature encoding processing, feature selection processing, data discretization processing, principal component analysis processing, and data enhancement processing. The specific processing is selected according to the data required by the target classification model.

[0047] Optionally, taking target object A as an example, based on the identity information of target object A, the original risk indicator data of target object A is obtained from the cloud server of a bank storage user; regional data - Beijing, industry data - manufacturing, historical credit limit data - good, credit score data - good, debt-to-income ratio data - good, credit record data - good, enterprise size data - medium, among which the enterprise size data in the original risk indicator data is data that is not needed by the target classification model and needs to be eliminated, and the regional data, industry data, credit score data, debt-to-income ratio data, credit record data, and enterprise size data need to be converted into numerical data.

[0048] This embodiment can eliminate the scale differences between different original risk indicator data and reduce the noise and abnormal values ​​in the original risk indicator data by performing target processing on the original risk indicator data, thereby achieving the purpose of improving the quality of the risk indicator data.

[0049] In an exemplary embodiment, before inputting the plurality of risk indicator data into the target classification model to obtain the target risk factor, the method further includes: acquiring the original training risk indicator data of the plurality of training objects from the target platform based on the identification information of the plurality of training objects, wherein the original training risk indicator data is used to describe the credit risk of the training objects, and each of the training objects includes one or more original training risk indicator data; performing target processing on the plurality of original training risk indicator data respectively to obtain the plurality of training risk indicator data, wherein the target processing is used to adjust the contribution of the original training risk indicator data to the classification sub-model; extracting the plurality of dimensions of the plurality of training risk indicator data based on the attribute information of the plurality of training risk indicator data, wherein each dimension includes one or more training risk indicator data; generating the multi-dimensional three-dimensional grid based on the plurality of dimensions and the plurality of training risk indicator data, wherein one training object corresponds to one grid in the multi-dimensional three-dimensional grid, and one grid is composed of the plurality of training risk indicators of different dimensions.

[0050] Optionally, the attribute information includes, but is not limited to, regional attributes, industry attributes, historical credit limit attributes, credit score attributes, debt-to-income ratio attributes, credit record attributes, and asset-liability attributes. Dimensions include, but are not limited to, region, industry, historical credit limit, debt-to-income ratio, credit score, credit record, and asset-liability.

[0051] Optionally, it is assumed that the training risk indicator data of training object A is obtained: region data-Beijing, industry data-manufacturing, historical credit line data-good, the training risk indicator data of training object B includes: region data-Tianjin, industry data-health care industry, historical credit line data-good, and the training risk indicator data of training object C includes: region data-Hebei, industry data-construction industry, historical credit line data-good, credit score data-excellent, and debt-to-income ratio data-average. Based on the attribute information of the above training risk indicator data: region attribute, industry attribute, historical credit line attribute, credit score attribute, and debt-to-income ratio data, the dimensions are extracted: region, industry, historical credit line, debt-to-income ratio, and credit score, wherein the training risk indicator data included in the region dimension are: Tianjin, Beijing, and Hebei, the training risk indicator data included in the industry dimension are: manufacturing, healthcare, and construction, the training risk indicator data included in the historical credit line are: 1 million, 500,000, and 600,000, the training risk indicator data included in the debt-to-income ratio are: average, and the training indicator data included in the credit score are: excellent. A five-dimensional multidimensional grid is generated based on the above dimensions and training risk indicator data, and the training objects A, B, and C belong to a grid in the above multidimensional grid.

[0052] This embodiment constructs a multi-dimensional three-dimensional grid through classification sub-models to comprehensively, accurately and efficiently evaluate the training risk indicators, thereby achieving the purpose of subsequently accurately determining the risk factors of the training objects.

[0053] In an exemplary embodiment, before inputting the above-mentioned multiple risk indicator data into the target classification model to obtain the target risk factor, the above-mentioned method also includes: obtaining the first original parameter, the second original parameter and the third original parameter of each of the above-mentioned training objects from the target platform based on the identification information of the above-mentioned multiple training objects, wherein the above-mentioned first original parameter is used to represent the original loan amount of the above-mentioned training object, the above-mentioned second original parameter is used to represent the original loan expenditure balance of the above-mentioned training object, and the above-mentioned third original parameter is used to represent the original non-performing loan balance of the above-mentioned training object; based on the initial training risk indicator coefficients of the above-mentioned multiple training risk indicator data of each training object in the above-mentioned multi-dimensional three-dimensional grid, calculating the initial risk factor of each of the above-mentioned training objects, wherein one of the above-mentioned training risk indicator data corresponds to one of the above-mentioned initial training risk indicator coefficients; calculating the product of the initial risk factor of each of the above-mentioned training objects and the above-mentioned first original parameter to obtain the first target parameter of each of the above-mentioned training objects, wherein the above-mentioned first target parameter is used to represent the initial predicted loan amount; based on The above-mentioned first original parameter, the above-mentioned second original parameter and the above-mentioned first target parameter are used to calculate the above-mentioned second target parameter of each of the above-mentioned training objects, wherein the above-mentioned second target parameter is used to represent the initial predicted loan expenditure balance; based on the above-mentioned second original parameter, the above-mentioned third original parameter and the above-mentioned second target parameter, the third target parameter is used to represent the initial predicted non-performing loan balance; the ratio of the first sum value to the second sum value is calculated to obtain the above-mentioned non-performing rate, wherein the above-mentioned first sum value is the sum value of the above-mentioned third target parameters corresponding to the above-mentioned multiple training objects, and the above-mentioned second sum value is the sum value of the above-mentioned second target parameters corresponding to the above-mentioned multiple training objects; under the condition of satisfying the above-mentioned constraints, the initial training risk indicator coefficient of the above-mentioned multiple training risk indicator data of each of the above-mentioned training objects in the above-mentioned multi-dimensional three-dimensional grid is adjusted to adjust the above-mentioned initial risk factor, the above-mentioned second target parameter and the above-mentioned third target parameter so that the above-mentioned non-performing rate is less than the above-mentioned first threshold value, and the risk factor of each of the above-mentioned training objects is obtained.

[0054] Optionally, training object A, training object B, and training object C are taken as examples for illustration.

[0055] First, based on the identity information of the training objects, the original loan amount of training object A in 2023 was obtained from a bank's database: 500,000 yuan, original loan expenditure balance of 500,000 yuan, and original non-performing loan balance of 50,000 yuan. The original loan amount of training object B in 2023 was 1,200,000 yuan, the original loan expenditure balance was 1,100,000 yuan, and the original non-performing loan balance was 50,000 yuan. The original loan amount of training object C in 2023 was 800,000 yuan, the original loan expenditure balance was 750,000 yuan, and the original non-performing loan balance was 100,000 yuan.

[0056] Then, based on the initial training risk indicator coefficients of the training risk indicator data of training objects A, B, and C in the multi-dimensional three-dimensional grid: training object A: Beijing -1.8, manufacturing industry -0.5, good historical credit line data -1.6, training object B: Tianjin -1.5, healthcare industry -1.0, general historical credit line data -0.5, training object C: Hebei -1.4, construction industry -0.4, good historical credit line data -1.6, excellent credit score data -0.8, general debt-to-income ratio data -0.5, the initial risk factor of training object A is calculated: 1.8×0.5×1.6=1.44, the initial risk factor of training object B: 1.5×1.0×0.5=0.75, and the initial risk factor of training object C: 1.4×0.4×1.6=0.896.

[0057] Then, the initial predicted loan amount is calculated based on the initial risk factor and the original loan amount: training object A-500,000×1.44=720,000 yuan, training object B-1,200,000×0.75=900,000 yuan, training object C-800,000×0.896=716,800 yuan.

[0058] Then, based on the original loan amount, original loan expenditure balance, original non-performing loan balance and initial predicted loan amount of training objects A, B, C, the initial predicted loan expenditure balance and initial predicted non-performing loan balance of training objects A, B, C are calculated.

[0059] Finally, based on the ratio of the initial predicted loan expenditure balance and value to the initial predicted non-performing loan balance and value of the training objects A, B, and C, the non-performing rate is obtained. Under the condition that the change range of the original loan amount and value of the training objects A, B, and C does not exceed +-10% and the initial training risk indicator coefficient index binning remains unchanged, the planning solution idea is adopted, and the Bayesian optimization algorithm and the Sequential Least Squares Quasi-Newton (SLSQP) algorithm are used to adjust the initial training risk indicator coefficients of multiple training risk indicator data of the training objects A, B, and C in the multi-dimensional three-dimensional grid, so as to adjust the initial risk factors, initial predicted loan expenditure balances, and initial predicted non-performing loan balances of the training objects A, B, and C, so that the non-performing rate is less than the first threshold, and the risk factors of the training objects A, B, and C at this time are obtained.

[0060] This embodiment adjusts the risk indicator coefficient of each training object by comprehensively considering the original loan amount, original loan expenditure balance, original non-performing loan balance, and non-performing rate, and evaluates the credit risk of the training object from a global perspective, thereby achieving the purpose of accurately determining the risk factor of the training object.

[0061] In an exemplary embodiment, based on the first original parameter, the second original parameter and the first target parameter, the second target parameter of each training object is calculated, including: when the first original parameter is equal to the second original parameter, the value of the first target parameter is determined as the second target parameter; when the first original parameter is greater than the second original parameter, the minimum value between the first target parameter and the second original parameter is determined as the second target parameter.

[0062] Optionally, training object A, training object B, and training object C are used as examples for illustration. The original loan amount of training object A is 500,000 yuan = the original loan expenditure balance is 500,000 yuan, so the initial predicted loan expenditure balance of training object A = the initial predicted loan amount of 720,000 yuan. The original loan amount of training object B is 1,200,000 yuan > the original loan expenditure balance is 1,100,000 yuan, so the initial predicted loan expenditure balance of training object B = min (initial predicted loan amount of 900,000 yuan, 1,100,000 yuan). The original loan amount of training object C is 800,000 yuan > the original loan expenditure balance is 750,000 yuan, so the initial predicted loan expenditure balance of training object C = min (initial predicted loan amount of 716,800 yuan, original loan expenditure balance of 750,000 yuan). This embodiment provides different calculation paths through this method, and flexibly determines the value of the second target parameter according to the relationship between the original parameters, thereby achieving the purpose of accurately determining the second target parameter.

[0063] In an exemplary embodiment, based on the second original parameter, the third original parameter and the second target parameter, the third target parameter of each training object is calculated, including: when the second original parameter is equal to the third original parameter, the value of the second target parameter is determined as the third target parameter; when the first difference between the second original parameter and the third original parameter is greater than or equal to the second target parameter, the third target parameter is determined to be zero; when the first difference is less than the second target parameter, the difference between the second target parameter and the first difference is determined as the third target parameter.

[0064] Optionally, training object A, training object B, and training object C are used as examples for illustration. The original loan expenditure balance of training object A is 500,000 yuan = the original non-performing loan balance is 50,000 yuan, so the initial predicted non-performing loan balance of training object A is 720,000 yuan. The difference between the original loan expenditure balance of training object B, 1,100,000 yuan and the original non-performing loan balance of 50,000 yuan is 1,050,000> the initial predicted loan expenditure balance of 900,000 yuan, so the initial predicted non-performing loan balance of training object B is 0. The difference between the original loan expenditure balance of training object C of 750,000 yuan and the original non-performing loan balance of 100,000 yuan is 650,000 yuan < the initial predicted loan expenditure balance of 716,800 yuan, so the initial predicted non-performing loan balance of training object C = the difference between the initial predicted loan expenditure balance of 716,800 yuan and 650,000 yuan, which is 66,800 yuan. This embodiment provides different calculation paths through this method, and flexibly determines the value of the third target parameter according to the relationship between the original parameters, thereby achieving the purpose of accurately determining the third target parameter.

[0065] In an exemplary embodiment, the target parameters of the target object are determined using the target risk factor, including: calculating the product of the target risk factor and the target original parameter of the target object to obtain the target parameter, wherein the target original parameter is used to represent the original loan amount of the target object.

[0066] Optionally, assuming that the final bad debt rate of training objects A, B, and C is less than the first threshold, the risk factors are 0.5, 0.8, and 1.1, respectively. Then, based on the final adjusted risk factors and the original loan amounts, the predicted loan amounts of training objects A, B, and C in 2024 are calculated.

[0067] The present invention will be described below in conjunction with specific embodiments:

[0068] Figure 3It is a flow chart of a method for determining a parameter in a specific embodiment of the present application, comprising the following steps:

[0069] S302, model design mainly includes: determining constraint conditions, building a multi-dimensional three-dimensional grid, and determining the defect rate;

[0070] Constraints: When the NPL ratio is the lowest, the constraints mainly include four categories: the original loan amount and value change range, the initial training risk indicator coefficient indicator bin adjustment range limit, the initial training risk indicator coefficient indicator bin monotonicity, and the initial training risk indicator coefficient indicator bin variability. Since the NPL ratio of different bins of each training risk indicator data is monotonic, from the perspective of business interpretation, the adjustment coefficients of different bins should also be monotonic. From the perspective of business interpretation, the adjustment coefficients of different bins of the same training risk indicator data should be both positive and negative;

[0071] Multi-dimensional grid: The training risk indicator data in the multi-dimensional grid reflects the risk of the training object from different dimensions;

[0072] NPL ratio: The purpose of model optimization is to minimize the NPL ratio (the ratio of the total predicted NPL balance to the total predicted loan expenditure balance). The optimization process is to adjust the initial training risk indicator coefficient of each training risk indicator data in the multi-dimensional grid, thereby adjusting the risk factor of each training object to be small, and adjusting the original credit limit of each training object to obtain the predicted loan limit, predicted expenditure balance, and predicted NPL balance, thereby obtaining the total predicted NPL balance and the total predicted expenditure balance.

[0073] Assume that the training risk indicator data of training object A is obtained: region data-Beijing, industry data-manufacturing, historical credit line data-good, the training risk indicator data of training object B includes: region data-Tianjin, industry data-health care industry, historical credit line data-good, and the training risk indicator data of training object C includes: region data-Hebei, industry data-construction industry, historical credit line data-good, credit score data-excellent, and debt-to-income ratio data-average. Based on the attribute information of the above training risk indicator data: region attribute, industry attribute, historical credit line attribute, credit score attribute, and debt-to-income ratio data, the dimensions are extracted: region, industry, historical credit line, debt-to-income ratio, and credit score, among which the training risk indicator data included in the region dimension are: Tianjin, Beijing, and Hebei, the training risk indicator data included in the industry dimension are: manufacturing, healthcare industry, and construction industry, the training risk indicator data included in the historical credit line are: 1 million, 500,000, and 600,000, the training risk indicator data included in the debt-to-income ratio are: average, and the training indicator data included in the credit score are: excellent. A five-dimensional multidimensional grid is generated based on the above dimensions and training risk indicator data, and the training objects A, B, and C belong to a grid in the above multidimensional grid.

[0074] Then, based on the identity information of the training object, the original loan amount of training object A in 2023 was obtained from the database of a certain bank, which was 500,000 yuan, the original loan expenditure balance was 500,000 yuan, and the original non-performing loan balance was 50,000 yuan. The original loan amount of training object B in 2023 was 1,200,000 yuan, the original loan expenditure balance was 1,100,000 yuan, and the original non-performing loan balance was 50,000 yuan. The original loan amount of training object C in 2023 was 800,000 yuan, the original loan expenditure balance was 750,000 yuan, and the original non-performing loan balance was 100,000 yuan. Then, based on the initial training risk indicator coefficients of the training risk indicator data of training objects A, B, and C in the multi-dimensional three-dimensional grid: training object A: Beijing -1.8, manufacturing industry -0.5, good historical credit line data -1.6, training object B: Tianjin -1.5, healthcare industry -1.0, general historical credit line data -0.5, training object C: Hebei -1.4, construction industry -0.4, good historical credit line data -1.6, excellent credit score data -0.8, general debt-to-income ratio data -0.5, the initial risk factor of training object A is calculated: 1.8×0.5×1.6=1.44, the initial risk factor of training object B: 1.5×1.0×0.5=0.75, and the initial risk factor of training object C: 1.4×0.4×1.6=0.896. Then, the initial predicted loan amount is calculated based on the initial risk factor and the original loan amount: training object A-500,000×1.44=720,000 yuan, training object B-1,200,000×0.75=900,000 yuan, training object C-800,000×0.896=716,800 yuan. Then, based on the original loan amount, original loan expenditure balance, original non-performing loan balance and initial predicted loan amount of training objects A, B, and C, the initial predicted loan expenditure balance and initial predicted non-performing loan balance of training objects A, B, and C are calculated. Finally, based on the ratio of the initial predicted loan expenditure balance and value of the training objects A, B, and C to the initial predicted non-performing loan balance and value, the non-performing rate is obtained. Under the condition that the change range of the original loan amount and value of the training objects A, B, and C does not exceed +-10% and the initial training risk indicator coefficient index bin remains unchanged, the planning solution idea is adopted, and the Bayesian optimization algorithm and the Sequential Least Squares Quasi-Newton (SLSQP) algorithm are used to adjust the initial training risk indicator coefficients of multiple training risk indicator data of the training objects A, B, and C in the multi-dimensional three-dimensional grid, so as to adjust the initial risk factors, initial predicted loan expenditure balances, and initial predicted non-performing loan balances of the training objects A, B, and C, so that the non-performing rate is less than the first threshold, and the risk factors of the training objects A, B, and C at this time are obtained;

[0075] Then, the original loan amount of training object A is 500,000 yuan = the original loan expenditure balance is 500,000 yuan, so the initial predicted loan expenditure balance of training object A = the initial predicted loan amount of 720,000 yuan. The original loan amount of training object B is 1,200,000 yuan > the original loan expenditure balance is 1,100,000 yuan, so the initial predicted loan expenditure balance of training object B = min (initial predicted loan amount 900,000 yuan, 1,100,000 yuan). The original loan amount of training object C is 800,000 yuan > the original loan expenditure balance is 750,000 yuan, so the initial predicted loan expenditure balance of training object C = min (initial predicted loan amount 716,800 yuan, original loan expenditure balance 750,000 yuan);

[0076] Then, the original loan expenditure balance of training object A is 500,000 yuan = the original non-performing loan balance of 50,000 yuan, so the initial predicted non-performing loan balance of training object A = the initial predicted loan expenditure balance of 720,000 yuan. The difference between the original loan expenditure balance of training object B, 1,100,000 yuan and the original non-performing loan balance of 50,000 yuan is 1,050,000> the initial predicted loan expenditure balance of 900,000 yuan, so the initial predicted non-performing loan balance of training object B = 0. The difference between the original loan expenditure balance of training object C, 750,000 yuan and the original non-performing loan balance of 100,000 yuan is 650,000 yuan < the initial predicted loan expenditure balance of 716,800 yuan, so the initial predicted non-performing loan balance of training object C = the difference between the initial predicted loan expenditure balance of 716,800 yuan and 650,000 yuan, 66,800 yuan;

[0077] Finally, assuming that the final bad debt rates of training objects A, B, and C are less than the first threshold, the risk factors are 0.5, 0.8, and 1.1, respectively. Then, based on the final adjusted risk factors and the original loan amounts, the predicted loan amounts of training objects A, B, and C in 2024 are calculated.

[0078] S304: In the process of model solving, based on the principles of high data quality, good interpretability and good model effect, the training risk indicator data, algorithm and constraints are continuously adjusted and optimized to finally determine the model solution. Considering different training risk indicator data, algorithm models and constraints, the idea of ​​operational optimization is applied to model building, and the model is iterated according to the model results to select the optimal credit line allocation solution;

[0079] S306, based on the optimal model results, analyzing the final risk factor, the distribution of the predicted loan amount, and the distribution of the change in the original loan amount, etc., to examine the stability and effectiveness of the model results;

[0080] S308, using the optimization principle to solve the initial training risk index coefficient index binning under the condition of the lowest bad rate, in order to ensure the generalization of the index binning adjustment coefficient, it is necessary to verify the stability of the index binning and the bad rate outside the sample and across periods. This embodiment uses the Population Stability Index (PSI) indicator to verify stability. PSI measures the stability of data by comparing the fluctuations of historical sample distribution and current sample distribution. It usually includes feature PSI and model PSI. Feature PSI focuses on whether the value of the feature fluctuates greatly over time, and can be used for model training and feature selection before going online, variable monitoring, etc. Model PSI focuses on whether the distribution of the model is stable after the training sample and model are deployed online. Unlike conventional machine learning models, this optimization model does not involve model PSI, but it is still necessary to perform stability testing on the feature variable PSI. The smaller the difference between the two samples, the smaller the PSI value, which means the more stable it is. Generally speaking, PSI<0.1, the sample distribution has a slight change, the variable is relatively stable and can be entered into the model, or the model is relatively stable and no adjustment is required. 0.1≤PSI<0.2, the sample distribution has changed, the variables fluctuate and should be carefully included in the model, or the model fluctuates, the cause of the change needs to be checked and the monitoring frequency needs to be strengthened. PSI>0.2, the sample distribution has changed significantly, the variables are not recommended to be included in the model, or the model needs to be adjusted.

[0081] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0082] In the present embodiment, a parameter determination device is also provided, and the device is used to implement the above-mentioned embodiment and preferred implementation mode, and the description has been made no further. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware is also possible and conceived.

[0083] Figure 4 is a structural block diagram of a parameter determination device according to an embodiment of the present application, such as Figure 4 As shown, the device comprises:

[0084] A first acquisition module 402 is used to acquire a plurality of risk indicator data of a target object, wherein the risk indicator data is used to describe the credit risk of the target object;

[0085] A first input module 404 is used to input the above-mentioned multiple risk indicator data into a target classification model to obtain a target risk factor, wherein the above-mentioned target classification model is obtained by training an initial classification model using multiple training risk indicator data of multiple training objects, the above-mentioned training objects include the above-mentioned target objects, and the above-mentioned initial classification model includes a classification sub-model and an adjustment sub-model, the above-mentioned classification sub-model is used to generate a multi-dimensional three-dimensional grid based on the above-mentioned multiple training risk indicator data, and the above-mentioned adjustment sub-model is used to adjust the initial training risk indicator coefficients of the multiple training risk indicator data in the above-mentioned multi-dimensional three-dimensional grid under the condition of satisfying constraint conditions, so that the defect rate of the multiple training objects is less than a first threshold value, and output a risk factor corresponding to each of the above-mentioned training objects based on the adjusted training risk indicator coefficient;

[0086] The first determination module 406 is used to determine the target parameters of the target object using the target risk factors.

[0087] In an exemplary embodiment, the above-mentioned first acquisition module includes: a first acquisition sub-module, used to obtain the identification information of the above-mentioned target object, wherein the above-mentioned identification information is information used to identify the above-mentioned target object; a second acquisition sub-module, used to obtain multiple original risk indicator data of the above-mentioned target object from the target platform based on the above-mentioned identification information, wherein the above-mentioned original risk indicator data is used to describe the credit risk of the above-mentioned target object; a first processing sub-module, used to perform target processing on the multiple original risk indicator data respectively to obtain the multiple risk indicator data, wherein the above-mentioned target processing is used to adjust the contribution of the above-mentioned risk indicator data to the above-mentioned target classification model.

[0088] In an exemplary embodiment, the device further includes: a second acquisition module, used to input the plurality of risk indicator data into a target classification model, and before obtaining a target risk factor, obtain the original training risk indicator data of the plurality of training objects from a target platform based on the identification information of the plurality of training objects, wherein the original training risk indicator data is used to describe the credit risk of the training object, and each of the training objects includes one or more original training risk indicator data; a first processing module, used to perform target processing on the plurality of original training risk indicator data respectively to obtain the plurality of training risk indicator data, wherein the target processing is used to adjust the contribution of the original training risk indicator data to the classification sub-model; a first extraction module, used to extract the plurality of dimensions of the plurality of training risk indicator data based on the attribute information of the plurality of training risk indicator data, wherein each dimension includes one or more training risk indicator data; a first generation module, used to generate the multi-dimensional three-dimensional grid based on the plurality of dimensions and the plurality of training risk indicator data, wherein one training object corresponds to one grid in the multi-dimensional three-dimensional grid, and one grid is composed of a plurality of training risk indicators of different dimensions.

[0089] In an exemplary embodiment, the device further includes: a third acquisition module, which is used to input the plurality of risk indicator data into the target classification model, and before obtaining the target risk factor, obtain the first original parameter, the second original parameter and the third original parameter of each of the training objects from the target platform based on the identification information of the plurality of training objects, wherein the first original parameter is used to represent the original loan amount of the training object, the second original parameter is used to represent the original loan expenditure balance of the training object, and the third original parameter is used to represent the original non-performing loan balance of the training object; a first calculation module, which is used to calculate the initial risk factor of each of the training objects based on the initial training risk indicator coefficient of the plurality of training risk indicator data of each training object in the multi-dimensional three-dimensional grid, wherein one training risk indicator data corresponds to one initial training risk indicator coefficient; a second calculation module, which is used to calculate the product of the initial risk factor of each of the training objects and the first original parameter, and obtain the first target parameter of each of the training objects, wherein the first target parameter is used to represent the initial predicted loan amount; a third calculation module, A fourth calculation module is used to calculate the third target parameter of each training object based on the first original parameter, the second original parameter and the first target parameter, wherein the second target parameter is used to represent the initial predicted loan expenditure balance; a fifth calculation module is used to calculate the ratio of the first sum value to the second sum value to obtain the non-performing rate, wherein the first sum value is the sum value of the multiple third target parameters corresponding to the multiple training objects, and the second sum value is the sum value of the multiple second target parameters corresponding to the multiple training objects; a first adjustment module is used to adjust the initial training risk indicator coefficient of the multiple training risk indicator data of each training object in the multi-dimensional three-dimensional grid under the condition of satisfying the above constraint conditions, so as to adjust the initial risk factor, the second target parameter and the third target parameter, so as to make the non-performing rate less than the first threshold value, and obtain the risk factor of each training object.

[0090] In an exemplary embodiment, the above-mentioned third calculation module includes: a first determination submodule, used to determine the value of the above-mentioned first target parameter as the above-mentioned second target parameter when the above-mentioned first original parameter is equal to the above-mentioned second original parameter; a second determination submodule, used to determine the minimum value of the above-mentioned first target parameter and the above-mentioned second original parameter as the above-mentioned second target parameter when the above-mentioned first original parameter is greater than the above-mentioned second original parameter.

[0091] In an exemplary embodiment, the fourth calculation module includes: a third determination submodule, used to determine the value of the second target parameter as the third target parameter when the second original parameter is equal to the third original parameter; a fourth determination submodule, used to determine the third target parameter as zero when the first difference between the second original parameter and the third original parameter is greater than or equal to the second target parameter; and a fifth determination submodule, used to determine the difference between the second target parameter and the first difference as the third target parameter when the first difference is less than the second target parameter.

[0092] In an exemplary embodiment, the first determination module includes: a sixth calculation submodule, used to calculate the product of the target risk factor and the target original parameter of the target object to obtain the target parameter, wherein the target original parameter is used to represent the original loan amount of the target object.

[0093] An embodiment of the present application provides a computer program product, including a computer program, and when the computer program is executed by a processor, the steps in any one of the above method embodiments are implemented.

[0094] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.

[0095] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0096] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0097] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0098] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.

[0099] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0100] The above are only preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for determining a parameter, characterized in that: include: Acquire multiple risk indicator data of a target object, wherein the risk indicator data is used to describe the credit risk of the target object; Inputting a plurality of the risk indicator data into a target classification model to obtain a target risk factor, wherein the target classification model is obtained by training an initial classification model using a plurality of training risk indicator data of a plurality of training objects, and the initial classification model comprises a classification sub-model and an adjustment sub-model, the classification sub-model is used to generate a multi-dimensional three-dimensional grid based on the plurality of the training risk indicator data, and the adjustment sub-model is used to adjust the initial training risk indicator coefficients of the plurality of the training risk indicator data in the multi-dimensional three-dimensional grid under the condition that a constraint condition is satisfied, so that the defect rate of the plurality of the training objects is less than a first threshold value, and output a risk factor corresponding to each of the training objects based on the adjusted training risk indicator coefficient; The target risk factor is used to determine the target parameter of the target object.

2. The method according to claim 1, characterized in that Obtain multiple risk indicator data of the target object, including: Acquire identification information of the target object, wherein the identification information is information used to identify the target object; Acquire multiple original risk indicator data of the target object from the target platform based on the identification information, wherein the original risk indicator data is used to describe the credit risk of the target object; Target processing is performed on the plurality of original risk indicator data respectively to obtain the plurality of risk indicator data, wherein the target processing is used to adjust the contribution of the risk indicator data to the target classification model.

3. The method according to claim 1, characterized in that Before inputting the plurality of risk indicator data into the target classification model to obtain the target risk factor, the method further comprises: Acquire original training risk indicator data of a plurality of training objects from a target platform based on identification information of the plurality of training objects, wherein the original training risk indicator data is used to describe the credit risk of the training object, and each training object includes one or more original training risk indicator data; Performing target processing on the plurality of original training risk indicator data respectively to obtain the plurality of training risk indicator data, wherein the target processing is used to adjust the contribution of the original training risk indicator data to the classification sub-model; Extracting multiple dimensions of the multiple training risk indicator data based on the attribute information of the multiple training risk indicator data, wherein each dimension includes one or more training risk indicator data; The multidimensional three-dimensional grid is generated based on the multiple dimensions and the multiple training risk indicator data, wherein one training object corresponds to a grid in the multidimensional three-dimensional grid, and one grid is composed of multiple training risk indicators of different dimensions.

4. The method according to claim 1, characterized in that: Before inputting the plurality of risk indicator data into the target classification model to obtain the target risk factor, the method further comprises: Based on the identification information of the plurality of training objects, a first original parameter, a second original parameter and a third original parameter of each training object are obtained from the target platform, wherein the first original parameter is used to represent the original loan amount of the training object, the second original parameter is used to represent the original loan expenditure balance of the training object, and the third original parameter is used to represent the original non-performing loan balance of the training object; Calculate the initial risk factor of each training object based on the initial training risk indicator coefficient of the plurality of training risk indicator data of each training object in the multi-dimensional three-dimensional grid, wherein one training risk indicator data corresponds to one initial training risk indicator coefficient; Calculating the product of the initial risk factor of each training object and the first original parameter to obtain a first target parameter of each training object, wherein the first target parameter is used to represent an initial predicted loan amount; Calculating the second target parameter of each training object based on the first original parameter, the second original parameter and the first target parameter, wherein the second target parameter is used to represent the initial predicted loan payment balance; Calculating a third target parameter of each training object based on the second original parameter, the third original parameter and the second target parameter, wherein the third target parameter is used to represent an initial predicted non-performing loan balance; Calculating a ratio of a first sum value to a second sum value to obtain the defect rate, wherein the first sum value is a sum value of a plurality of the third target parameters corresponding to a plurality of the training objects, and the second sum value is a sum value of a plurality of the second target parameters corresponding to a plurality of the training objects; Under the condition of satisfying the constraints, the initial training risk indicator coefficients of the multiple training risk indicator data of each training object in the multidimensional three-dimensional grid are adjusted to adjust the initial risk factor, the second target parameter and the third target parameter so that the defect rate is less than the first threshold, thereby obtaining the risk factor of each training object.

5. The method according to claim 4, characterized in that Calculating the second target parameter of each training object based on the first original parameter, the second original parameter and the first target parameter includes: When the first original parameter is equal to the second original parameter, determining the value of the first target parameter as the second target parameter; When the first original parameter is greater than the second original parameter, the minimum value of the first target parameter and the second original parameter is determined as the second target parameter.

6. The method according to claim 4, characterized in that Calculating a third target parameter of each training object based on the second original parameter, the third original parameter and the second target parameter includes: When the second original parameter is equal to the third original parameter, determining the value of the second target parameter as the third target parameter; When a first difference between the second original parameter and the third original parameter is greater than or equal to the second target parameter, determining the third target parameter to be zero; When the first difference is smaller than the second target parameter, the difference between the second target parameter and the first difference is determined as the third target parameter.

7. The method according to claim 1, characterized in that Determining the target parameters of the target object using the target risk factor includes: The product of the target risk factor and the target original parameter of the target object is calculated to obtain the target parameter, wherein the target original parameter is used to represent the original loan amount of the target object.

8. A parameter determination device, characterized in that: include: A first acquisition module is used to acquire a plurality of risk indicator data of a target object, wherein the risk indicator data is used to describe the credit risk of the target object; A first input module is used to input the plurality of risk indicator data into a target classification model to obtain a target risk factor, wherein the target classification model is obtained by training an initial classification model using a plurality of training risk indicator data of a plurality of training objects, and the initial classification model includes a classification sub-model and an adjustment sub-model, the classification sub-model is used to generate a multi-dimensional three-dimensional grid based on the plurality of training risk indicator data, the adjustment sub-model is used to adjust the initial training risk indicator coefficients of the plurality of training risk indicator data in the multi-dimensional three-dimensional grid under the condition of satisfying constraint conditions, so that the defect rate of the plurality of training objects is less than a first threshold value, and output a risk factor corresponding to each of the training objects based on the adjusted training risk indicator coefficient; The first determination module is used to determine the target parameter of the target object by using the target risk factor.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program implements the steps of the method described in any one of claims 1 to 7 when executed by a processor.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method described in any one of claims 1 to 7 are implemented.

Images