Enterprise credit risk assessment processing method and device

Abstract

The application provides an enterprise credit risk assessment processing method and device, relates to the technical field of risk assessment, and can be applied to the financial field or other technical fields.The method comprises the following steps: obtaining enterprise related data, preprocessing the enterprise related data to obtain initial feature data, performing feature engineering processing on the initial feature data to obtain feature data which has a dominant influence on enterprise credit risk assessment, performing credit risk assessment on the feature data based on a preset credit scoring card model to obtain a first scoring result, performing credit risk assessment on the fusion feature data based on a preset credit scoring model to obtain a second scoring result, and obtaining a total scoring result according to the first scoring result, the second scoring result and self-adaptive mediation hyperparameters.The device executes the above method.The method and device provided in the application embodiment can improve the applicability and accuracy of enterprise credit risk assessment.

Description

Technical Field

[0001] This invention relates to the field of risk assessment technology, specifically to a method and apparatus for corporate credit risk assessment. Background Technology

[0002] Current automated corporate credit risk assessment models often rely on specific scenarios for credit authorization, such as "tax source loans" and "settlement loans." Because these models have a single source of features, they are prone to granting higher credit to customers with higher overall risk, while also overlooking high-quality customers with slight deficiencies in certain aspects. Summary of the Invention

[0003] To address the problems in the prior art, embodiments of the present invention provide a method and apparatus for enterprise credit risk assessment, which can at least partially solve the problems existing in the prior art.

[0004] On the one hand, this invention proposes a method for corporate credit risk assessment, including:

[0005] Acquire relevant enterprise data, preprocess the relevant enterprise data, and obtain initial feature data;

[0006] The initial feature data is subjected to feature engineering to obtain feature data that has a significant impact on corporate credit risk assessment;

[0007] The feature data is assessed for credit risk based on a preset credit scoring card model to obtain a first scoring result, and the fused feature data is assessed for credit risk based on a preset credit scoring model to obtain a second scoring result.

[0008] The fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained based on the feature parameters and the weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data.

[0009] The total score is calculated based on the first score, the second score, and the hyperparameters of the adaptive adjustment; the hyperparameters of the adaptive adjustment are obtained from the evaluation index parameters of the model prediction results of the machine learning model.

[0010] The step of calculating the total score based on the first score result, the second score result, and the hyperparameters of adaptive adjustment includes:

[0011] The difference between 1 and the square of the adaptively adjusted hyperparameter is used as the first weight value of the first scoring result, and the square of the adaptively adjusted hyperparameter is used as the second weight value of the second scoring result.

[0012] The total score is obtained based on the first weight value, the first score result, the second weight value, and the second score result.

[0013] The evaluation index parameters for the model prediction results include the area under the receiver operating characteristic curve and the coordinate axis; correspondingly, the hyperparameters of the adaptive modulation are obtained based on the evaluation index parameters for the model prediction results of the machine learning model, including:

[0014] The area corresponding value is assigned to the hyperparameter of the adaptive adjustment.

[0015] The enterprise credit risk assessment and processing method further includes:

[0016] Within a preset initial period for credit risk assessment of fused feature data based on a preset credit scoring model, the hyperparameters of the adaptive adjustment are gradually increased from zero, and the hyperparameters of the adaptive adjustment are no longer increased when they reach a preset amplitude value.

[0017] When the preset initial time period is reached, the step of assigning the area corresponding value to the hyperparameter of the adaptive adjustment is performed.

[0018] The machine learning model includes at least one or more of the following: decision tree model, optimized distributed gradient boosting library, logistic regression model, and long short-term memory network model.

[0019] The enterprise credit risk assessment and processing method further includes:

[0020] Generate a credit analysis report and visualize the feature data and the credit analysis report.

[0021] The enterprise credit risk assessment and processing method further includes:

[0022] The system responds to the credit loan results obtained by the user based on the feature data and the credit analysis report, and updates the enterprise-related sample data in the training set based on the credit loan results.

[0023] On the one hand, the present invention proposes a corporate credit risk assessment and processing device, comprising:

[0024] The acquisition unit is used to acquire enterprise-related data, preprocess the enterprise-related data, and obtain initial feature data;

[0025] The processing unit is used to perform feature engineering on the initial feature data to obtain feature data that has a significant impact on enterprise credit risk assessment.

[0026] The evaluation unit is used to perform credit risk assessment on the feature data based on a preset credit scoring card model to obtain a first scoring result, and to perform credit risk assessment on the fused feature data based on a preset credit scoring model to obtain a second scoring result.

[0027] The fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained based on the feature parameters and the weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data.

[0028] The calculation unit is used to calculate the total score result based on the first score result, the second score result, and the adaptive adjustment hyperparameters; the adaptive adjustment hyperparameters are obtained based on the evaluation index parameters of the model prediction results of the machine learning model.

[0029] In another aspect, embodiments of the present invention provide an electronic device, including: a processor, a memory, and a bus, wherein,

[0030] The processor and the memory communicate with each other via the bus;

[0031] The memory stores program instructions that can be executed by the processor, and the processor can execute the following methods by calling the program instructions:

[0032] Acquire relevant enterprise data, preprocess the relevant enterprise data, and obtain initial feature data;

[0033] The initial feature data is subjected to feature engineering to obtain feature data that has a significant impact on corporate credit risk assessment;

[0034] The feature data is assessed for credit risk based on a preset credit scoring card model to obtain a first scoring result, and the fused feature data is assessed for credit risk based on a preset credit scoring model to obtain a second scoring result.

[0035] The fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained based on the feature parameters and the weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data.

[0036] The total score is calculated based on the first score, the second score, and the hyperparameters of the adaptive adjustment; the hyperparameters of the adaptive adjustment are obtained from the evaluation index parameters of the model prediction results of the machine learning model.

[0037] This invention provides a non-transitory computer-readable storage medium, comprising:

[0038] The non-transitory computer-readable storage medium stores computer instructions that cause the computer to perform the following methods:

[0039] Acquire relevant enterprise data, preprocess the relevant enterprise data, and obtain initial feature data;

[0040] The initial feature data is subjected to feature engineering to obtain feature data that has a significant impact on corporate credit risk assessment;

[0041] The feature data is assessed for credit risk based on a preset credit scoring card model to obtain a first scoring result, and the fused feature data is assessed for credit risk based on a preset credit scoring model to obtain a second scoring result.

[0042] The fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained based on the feature parameters and the weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data.

[0043] The total score is calculated based on the first score, the second score, and the hyperparameters of the adaptive adjustment; the hyperparameters of the adaptive adjustment are obtained from the evaluation index parameters of the model prediction results of the machine learning model.

[0044] The enterprise credit risk assessment processing method and apparatus provided in this invention acquires enterprise-related data, preprocesses the enterprise-related data to obtain initial feature data, performs feature engineering processing on the initial feature data to obtain feature data with a significant impact on enterprise credit risk assessment, performs credit risk assessment on the feature data based on a preset credit scoring card model to obtain a first scoring result, and performs credit risk assessment on fused feature data based on a preset credit scoring model to obtain a second scoring result; wherein, the fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained based on the feature parameters and weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data; a total scoring result is obtained based on the first scoring result, the second scoring result, and adaptive adjustment hyperparameters; the adaptive adjustment hyperparameters are obtained based on the evaluation index parameters of the model prediction results of the machine learning model, which can improve the applicability and accuracy of enterprise credit risk assessment. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0046] Figure 1 This is a flowchart illustrating a corporate credit risk assessment and processing method provided in an embodiment of the present invention.

[0047] Figure 2 This is a flowchart illustrating a corporate credit risk assessment and processing method provided in another embodiment of the present invention.

[0048] Figure 3 This is a flowchart illustrating a corporate credit risk assessment and processing method provided in another embodiment of the present invention.

[0049] Figure 4 This is a schematic diagram of the structure of an enterprise credit risk assessment and processing device provided in an embodiment of the present invention.

[0050] Figure 5 This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments and descriptions of the present invention are used to explain the present invention, but are not intended to limit the present invention. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be arbitrarily combined with each other.

[0052] Figure 1 This is a flowchart illustrating a corporate credit risk assessment and processing method according to an embodiment of the present invention, as shown below. Figure 1 As shown, the enterprise credit risk assessment method provided in this embodiment of the invention includes:

[0053] Step S1: Obtain relevant enterprise data, preprocess the relevant enterprise data, and obtain initial feature data.

[0054] Step S2: Perform feature engineering on the initial feature data to obtain feature data that has a significant impact on corporate credit risk assessment.

[0055] Step S3: Perform credit risk assessment on the feature data based on the preset credit scoring card model to obtain a first scoring result, and perform credit risk assessment on the fused feature data based on the preset credit scoring model to obtain a second scoring result;

[0056] The fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained based on the feature parameters and the weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data.

[0057] Step S4: Calculate the total score based on the first score result, the second score result, and the hyperparameters of the adaptive adjustment; the hyperparameters of the adaptive adjustment are obtained based on the evaluation index parameters of the model prediction results of the machine learning model.

[0058] In step S1 above, the device acquires enterprise-related data, preprocesses the enterprise-related data, and obtains initial feature data. The device can be a computer device that executes the method, such as a server. It should be noted that the acquisition and analysis of data involved in this embodiment of the invention are authorized by the user. Enterprise-related data can specifically include enterprise-related data of micro and small enterprises. Micro and small enterprises refer to enterprises engaged in non-restricted and prohibited industries that simultaneously meet the following three conditions: annual taxable income not exceeding 3 million yuan, number of employees not exceeding 300, and total assets not exceeding 50 million yuan.

[0059] like Figure 2 As shown, enterprise-related data can include enterprise operating data, as well as physical assets owned by the enterprise, including data related to production equipment, factory buildings, and consumables. This data can be used as collateral for the enterprise.

[0060] The aforementioned enterprise-related data includes, but is not limited to, the enterprise's annual financial statements, operating settlement records, upstream and downstream transaction contracts, historical credit information, annual tax payment information, business registration information, litigation information, and public opinion news. In this process, the data can be divided into two parts: one part is provided by the borrower, and the other part is accessed from external network data.

[0061] Preprocessing of enterprise-related data, including data cleaning and missing value imputation, yields data that conforms to automated feature engineering processing specifications, i.e., initial feature data X. i X i It consists of n-dimensional feature data, that is, X i = (x1, x2, ..., x n ), where x i For X i Feature data in one dimension.

[0062] In step S2 above, the device performs feature engineering on the initial feature data to obtain feature data that has a significant impact on corporate credit risk assessment.

[0063] Based on expert experience, the data types of the aforementioned enterprise-related data can be analyzed using multiple preset feature engineering modules to process the initial feature data X. i Automatic feature extraction and feature filtering can yield m-dimensional feature data Y after feature engineering. i That is, Y i =(y1,y2...y m ), where y i For Y i One-dimensional feature data. In the feature engineering process, this method corresponds to different feature engineering modules for different types of original fields. The features processed by feature engineering are all explicit features, which have a significant impact on enterprise credit risk assessment and have clear field meanings, making them easy for credit personnel to access and review. For example, some common feature fields after processing are shown in Table 1:

[0064] Table 1

[0065]

[0066]

[0067] In step S3 above, the device performs credit risk assessment on the feature data based on a preset credit scoring card model to obtain a first scoring result, and performs credit risk assessment on the fused feature data based on the preset credit scoring model to obtain a second scoring result.

[0068] The fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained based on the feature parameters and the weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data.

[0069] The feature data Y i Input the pre-set credit scoring card model for scoring, and output the first scoring result, i.e., the scoring card result S. card The preset credit scoring card model used in this invention is based on the credit experts' experience in setting weight parameters for feature Y obtained during feature engineering. i By manually setting weight parameters Then we have S card =α1y1+α2y2+...+α m y m Since such scorecard models have a relatively simple structure and cannot fully explore the relationships between different features, this method also introduces a machine learning model based on large-scale data to improve the accuracy of the scoring.

[0070] Original feature X i and the feature Y output after feature engineering i By performing a fusion, or splicing operation, the fused feature Z can be obtained. i Z can then be obtained. i = (x1, x2, ..., x n y1, y2, ..., y m ), to Z i The input is fed into a preset credit scoring model to obtain the second scoring result, i.e., the scoring result S. model The S obtained above card and S model The range of values ​​is limited to the interval [0,1].

[0071] Based on the existing corporate credit risk database, it is divided into a training set and a test set, namely: Using the training set Train the machine learning model and test it on the test set. The training results can be validated using the model prediction results and evaluation metrics to determine the validation results.

[0072] The evaluation index parameter for the above model prediction results can be the area under the receiver operating characteristic (ROC) curve and the coordinate axes (AUC). The AUC value can be calculated as follows:

[0073]

[0074] in, Ins represents the rank of the i-th sample's index. i Let represent the index of the i-th sample, and M and N be the number of positive and negative samples, respectively. This AUC value is set as the adaptive adjustment hyperparameter μ, i.e., μ = AUC. In practical applications, the machine learning algorithms used in this method include one or more of the following: decision tree model, optimized distributed gradient boosting library, logistic regression model, and long short-term memory network model. That is, the machine learning algorithm can be a single model or a combined model based on the above single models.

[0075] In step S4 above, the device calculates the total score based on the first score result, the second score result, and the adaptive adjustment hyperparameters; the adaptive adjustment hyperparameters are obtained based on the evaluation index parameters of the model prediction results of the machine learning model.

[0076] The step of calculating the total score based on the first score result, the second score result, and the adaptively adjusted hyperparameters includes:

[0077] The difference between 1 and the square of the adaptively adjusted hyperparameter is used as the first weight value of the first scoring result, and the square of the adaptively adjusted hyperparameter is used as the second weight value of the second scoring result.

[0078] The total score is obtained based on the first weight value, the first scoring result, the second weight value, and the second scoring result. The total score can be obtained using the following formula:

[0079] S i =(1-μ 2 )S card +μ 2 S model

[0080] It should be noted that, in the initial stage of using the preset credit scoring model, there may be insufficient data. In order to ensure the accuracy of the overall score, the initial score obtained based on the preset credit scoring model should be relied upon more during this period. After this period, when the data is sufficient, the overall score can be calculated according to the above formula.

[0081] like Figure 3 As shown, the enterprise credit risk assessment and processing method further includes:

[0082] Step R1: Within a preset initial period for credit risk assessment of fused feature data based on a preset credit scoring model, gradually increase the hyperparameters of the adaptive adjustment from zero, and stop increasing the hyperparameters of the adaptive adjustment when they reach a preset amplitude value. The preset initial period can be set independently according to the actual situation, or it can be determined according to the amount of data in the training set. If there is a lot of data, a small value can be selected for the preset initial period; if there is a little data, a large value can be selected for the preset initial period.

[0083] The preset amplitude value can also be set independently according to the actual situation to ensure that the upper limit of the hyperparameter of the adaptive adjustment does not exceed the preset amplitude value, thereby controlling the overall score result to rely more on the first score result obtained based on the preset credit scoring card model.

[0084] For example, the initial preset period can be set to one week, or 7 days, and the hyperparameter of the adaptive adjustment is increased once each day. The hyperparameter of the adaptive adjustment is adjusted to 0.02 on the first day, to 0.04 on the second day, and so on. If the preset amplitude value is 0.1, then when the hyperparameter of the adaptive adjustment is adjusted to 0.1, the hyperparameter of the adaptive adjustment will no longer be increased.

[0085] That is, gradually increasing the hyperparameters of the adaptive adjustment from zero can specifically include:

[0086] The hyperparameter of the adaptive adjustment is gradually increased from zero according to a preset increment value. The preset increment value can be set independently according to the actual situation and can be selected as 0.02.

[0087] Step R2: Upon reaching the preset initial time period, execute the step of assigning the area-corresponding value to the hyperparameter of the adaptive adjustment. Referring to the example above, starting from day 8, the AUC value obtained through the AUC value calculation method is assigned to the hyperparameter of the adaptive adjustment.

[0088] The enterprise credit risk assessment and processing method also includes:

[0089] Generate a credit analysis report and visualize the feature data and the credit analysis report for users, such as credit specialists, to view.

[0090] The enterprise credit risk assessment and processing method also includes:

[0091] The system responds to the credit loan results obtained by the user based on the feature data and the credit analysis report, and updates the enterprise-related sample data in the training set based on the credit loan results.

[0092] Users will share enterprise data and loan results {X i label i} should be added to the dataset in a timely manner. Since the success of loan repayment may not be known until a considerable period has passed, X needs to be... i Loan result label i Track and update the above label. i To label the loan outcome.

[0093] Whenever there is new data for {X} i label i Once entered into the database, the hyperparameters of the adaptive adjustment are updated adaptively through the AUC value calculation method described above, thereby continuously improving the accuracy of credit risk assessment.

[0094] The advantages of this invention are as follows:

[0095] (1) Feature extraction was performed on multiple information of micro and small enterprises, which has higher robustness.

[0096] (2) The above firstly enables credit personnel to quickly and comprehensively understand the company's operating conditions through automated feature engineering, automated visualization analysis and report generation, and can be mutually verified with the credit risk score results output by the model, reducing the possibility of model misjudgment in special circumstances.

[0097] (3) At the beginning, the above model can be made to rely more on the results of the scorecard based on expert experience by adjusting the adaptive parameters when the amount of data is insufficient or the accuracy of the machine learning model is low. At the same time, as the amount of data continues to accumulate, it can also achieve scoring accuracy that surpasses human judgment. Therefore, this method has a wide range of applications and room for growth, and can be deployed in financial institutions of different sizes and at different stages of development.

[0098] The enterprise credit risk assessment processing method provided in this invention involves acquiring relevant enterprise data, preprocessing the relevant enterprise data to obtain initial feature data, performing feature engineering on the initial feature data to obtain feature data that has a significant impact on enterprise credit risk assessment, conducting credit risk assessment on the feature data based on a preset credit scoring card model to obtain a first scoring result, and conducting credit risk assessment on fused feature data based on a preset credit scoring model to obtain a second scoring result. The fused feature data is obtained by fusing the initial feature data and the feature data. The preset credit scoring card model is obtained based on the feature parameters and weight parameters corresponding to the feature data. The preset credit scoring model is obtained by training a machine learning model based on relevant enterprise sample data. A total scoring result is calculated based on the first scoring result, the second scoring result, and adaptive adjustment hyperparameters. The adaptive adjustment hyperparameters are obtained based on the evaluation index parameters of the model prediction results of the machine learning model, which can improve the applicability and accuracy of enterprise credit risk assessment.

[0099] Further, the step of calculating the total score based on the first scoring result, the second scoring result, and the adaptively adjusted hyperparameters includes:

[0100] The difference between 1 and the square of the adaptively adjusted hyperparameter is used as the first weight value of the first scoring result, and the square of the adaptively adjusted hyperparameter is used as the second weight value of the second scoring result; the above description is provided and will not be repeated here.

[0101] The total score is obtained based on the first weight value, the first scoring result, the second weight value, and the second scoring result. This can be referred to the above explanation and will not be repeated here.

[0102] The enterprise credit risk assessment method provided in this embodiment of the invention can further improve the applicability and accuracy of enterprise credit risk assessment.

[0103] Further, the evaluation index parameters of the model prediction results include the area under the receiver operating characteristic curve and the coordinate axis; correspondingly, the hyperparameters of the adaptive modulation are obtained based on the evaluation index parameters of the model prediction results of the machine learning model, including:

[0104] The area corresponding value is assigned to the hyperparameter of the adaptive adjustment. This can be referred to the above explanation and will not be repeated here.

[0105] The enterprise credit risk assessment processing method provided in this invention reasonably determines the value of the hyperparameters for adaptive adjustment, thereby further improving the applicability and accuracy of enterprise credit risk assessment.

[0106] Furthermore, the enterprise credit risk assessment and processing method also includes:

[0107] Within a preset initial period for credit risk assessment of fused feature data based on a preset credit scoring model, the hyperparameters of the adaptive adjustment are gradually increased from zero, and the hyperparameters of the adaptive adjustment are no longer increased when they reach a preset amplitude value; this can be referred to the above description and will not be repeated here.

[0108] When the preset initial time period is reached, the step of assigning the area-corresponding value to the hyperparameter of the adaptive adjustment is executed. This can be referred to the above description and will not be repeated here.

[0109] The enterprise credit risk assessment processing method provided in this invention can further improve the applicability and accuracy of enterprise credit risk assessment by optimizing and adjusting the hyperparameters of adaptive adjustment at different time stages.

[0110] Furthermore, the machine learning model includes at least one or more of the following: decision tree model, optimized distributed gradient boosting library, logistic regression model, and long short-term memory network model. Refer to the above description; further details are omitted.

[0111] The enterprise credit risk assessment processing method provided in this embodiment of the invention can improve the accuracy of the second scoring result, and further improve the applicability and accuracy of enterprise credit risk assessment.

[0112] Furthermore, the enterprise credit risk assessment and processing method also includes:

[0113] Generate a credit analysis report and visualize the feature data and the credit analysis report. Refer to the above explanation; further details are omitted.

[0114] The enterprise credit risk assessment method provided in this embodiment of the invention makes it convenient for users to conduct credit risk assessments based on the displayed content.

[0115] Furthermore, the enterprise credit risk assessment and processing method also includes:

[0116] The system responds to the user's credit loan results obtained based on the feature data and the credit analysis report, and updates the relevant sample data of the enterprise in the training set according to the credit loan results. This can be referred to the above description and will not be repeated here.

[0117] The enterprise credit risk assessment processing method provided in this embodiment of the invention can further improve the applicability and accuracy of enterprise credit risk assessment by updating the relevant sample data of enterprises in the training set.

[0118] It should be noted that the enterprise credit risk assessment and processing method provided in the embodiments of the present invention can be used in the financial field, or in any technical field other than the financial field. The embodiments of the present invention do not limit the application field of the enterprise credit risk assessment and processing method.

[0119] Figure 4 This is a schematic diagram of the structure of an enterprise credit risk assessment and processing device provided in an embodiment of the present invention, as shown below. Figure 4 As shown, the enterprise credit risk assessment and processing device provided in this embodiment of the invention includes an acquisition unit 401, a processing unit 402, an assessment unit 403, and a calculation unit 404, wherein:

[0120] The acquisition unit 401 is used to acquire enterprise-related data and preprocess the enterprise-related data to obtain initial feature data; the processing unit 402 is used to perform feature engineering processing on the initial feature data to obtain feature data that has a significant impact on enterprise credit risk assessment; the evaluation unit 403 is used to perform credit risk assessment on the feature data based on a preset credit scoring card model to obtain a first scoring result, and to perform credit risk assessment on the fused feature data based on a preset credit scoring model to obtain a second scoring result; wherein, the fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained according to the feature parameters and the weight parameters corresponding to the feature data respectively; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data; the calculation unit 404 is used to calculate the total scoring result according to the first scoring result, the second scoring result and the adaptive adjustment hyperparameters; the adaptive adjustment hyperparameters are obtained according to the evaluation index parameters of the model prediction results of the machine learning model.

[0121] Specifically, the acquisition unit 401 in the device is used to acquire enterprise-related data, preprocess the enterprise-related data to obtain initial feature data; the processing unit 402 is used to perform feature engineering processing on the initial feature data to obtain feature data that has a significant impact on enterprise credit risk assessment; the evaluation unit 403 is used to perform credit risk assessment on the feature data based on a preset credit scoring card model to obtain a first scoring result, and to perform credit risk assessment on the fused feature data based on a preset credit scoring model to obtain a second scoring result; wherein, the fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained according to the feature parameters and the weight parameters corresponding to the feature data respectively; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data; the calculation unit 404 is used to calculate the total scoring result according to the first scoring result, the second scoring result and the adaptive adjustment hyperparameters; the adaptive adjustment hyperparameters are obtained according to the evaluation index parameters of the model prediction results of the machine learning model.

[0122] The enterprise credit risk assessment processing device provided in this embodiment of the invention acquires enterprise-related data, preprocesses the enterprise-related data to obtain initial feature data, performs feature engineering on the initial feature data to obtain feature data with a significant impact on enterprise credit risk assessment, performs credit risk assessment on the feature data based on a preset credit scoring card model to obtain a first scoring result, and performs credit risk assessment on fused feature data based on a preset credit scoring model to obtain a second scoring result; wherein, the fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained based on the feature parameters and weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data; a total scoring result is obtained based on the first scoring result, the second scoring result, and adaptive adjustment hyperparameters; the adaptive adjustment hyperparameters are obtained based on the evaluation index parameters of the model prediction results of the machine learning model, which can improve the applicability and accuracy of enterprise credit risk assessment.

[0123] Furthermore, the computing unit 404 is specifically used for:

[0124] The difference between 1 and the square of the adaptively adjusted hyperparameter is used as the first weight value of the first scoring result, and the square of the adaptively adjusted hyperparameter is used as the second weight value of the second scoring result.

[0125] The total score is obtained based on the first weight value, the first score result, the second weight value, and the second score result.

[0126] The enterprise credit risk assessment processing device provided in this embodiment of the invention can further improve the applicability and accuracy of enterprise credit risk assessment.

[0127] Furthermore, the evaluation index parameters of the model prediction results include the area under the receiver operating characteristic curve and the coordinate axis; correspondingly, the enterprise credit risk assessment and processing device is also used for:

[0128] The area corresponding value is assigned to the hyperparameter of the adaptive adjustment.

[0129] The enterprise credit risk assessment processing device provided in this embodiment of the invention can reasonably determine the value of the hyperparameter of adaptive adjustment, thereby further improving the applicability and accuracy of enterprise credit risk assessment.

[0130] Furthermore, the enterprise credit risk assessment and processing device is also used for:

[0131] Within a preset initial period for credit risk assessment of fused feature data based on a preset credit scoring model, the hyperparameters of the adaptive adjustment are gradually increased from zero, and the hyperparameters of the adaptive adjustment are no longer increased when they reach a preset amplitude value.

[0132] When the preset initial time period is reached, the step of assigning the area corresponding value to the hyperparameter of the adaptive adjustment is performed.

[0133] The enterprise credit risk assessment processing device provided in this embodiment of the invention can further improve the applicability and accuracy of enterprise credit risk assessment by optimizing and adjusting the hyperparameters of adaptive adjustment at different time stages.

[0134] Furthermore, the machine learning model includes at least one or more of the following: decision tree model, optimized distributed gradient boosting library, logistic regression model, and long short-term memory network model.

[0135] The enterprise credit risk assessment processing device provided in this embodiment of the invention can improve the accuracy of the second scoring result, and further improve the applicability and accuracy of enterprise credit risk assessment.

[0136] Furthermore, the enterprise credit risk assessment and processing device is also used for:

[0137] Generate a credit analysis report and visualize the feature data and the credit analysis report.

[0138] The enterprise credit risk assessment processing device provided in this embodiment of the invention facilitates users to conduct credit risk assessments based on the displayed content.

[0139] Furthermore, the enterprise credit risk assessment and processing device is also used for:

[0140] The system responds to the credit loan results obtained by the user based on the feature data and the credit analysis report, and updates the enterprise-related sample data in the training set based on the credit loan results.

[0141] The enterprise credit risk assessment processing device provided in this embodiment of the invention can further improve the applicability and accuracy of enterprise credit risk assessment by updating enterprise-related sample data in the training set.

[0142] The embodiments of the present invention provide an enterprise credit risk assessment and processing device that can be used to execute the processing flow of the above-described method embodiments. Its functions will not be repeated here, but can be referred to the detailed description of the above-described method embodiments.

[0143] Figure 5 This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention, such as... Figure 5 As shown, the electronic device includes: a processor 501, a memory 502, and a bus 503;

[0144] The processor 501 and the memory 502 communicate with each other via the bus 503.

[0145] The processor 501 is used to call program instructions in the memory 502 to execute the methods provided in the above-described method embodiments, including, for example:

[0146] Acquire relevant enterprise data, preprocess the relevant enterprise data, and obtain initial feature data;

[0147] The initial feature data is subjected to feature engineering to obtain feature data that has a significant impact on corporate credit risk assessment;

[0148] The feature data is assessed for credit risk based on a preset credit scoring card model to obtain a first scoring result, and the fused feature data is assessed for credit risk based on a preset credit scoring model to obtain a second scoring result.

[0149] The fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained based on the feature parameters and the weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data.

[0150] The total score is calculated based on the first score, the second score, and the hyperparameters of the adaptive adjustment; the hyperparameters of the adaptive adjustment are obtained from the evaluation index parameters of the model prediction results of the machine learning model.

[0151] This embodiment discloses a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, and when the program instructions are executed by a computer, the computer can perform the methods provided in the above-described method embodiments, such as:

[0152] Acquire relevant enterprise data, preprocess the relevant enterprise data, and obtain initial feature data;

[0153] The initial feature data is subjected to feature engineering to obtain feature data that has a significant impact on corporate credit risk assessment;

[0154] The feature data is assessed for credit risk based on a preset credit scoring card model to obtain a first scoring result, and the fused feature data is assessed for credit risk based on a preset credit scoring model to obtain a second scoring result.

[0155] The fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained based on the feature parameters and the weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data.

[0156] The total score is calculated based on the first score, the second score, and the hyperparameters of the adaptive adjustment; the hyperparameters of the adaptive adjustment are obtained from the evaluation index parameters of the model prediction results of the machine learning model.

[0157] This embodiment provides a computer-readable storage medium storing a computer program that causes the computer to execute the methods provided in the above-described method embodiments, including, for example:

[0158] Acquire relevant enterprise data, preprocess the relevant enterprise data, and obtain initial feature data;

[0159] The initial feature data is subjected to feature engineering to obtain feature data that has a significant impact on corporate credit risk assessment;

[0160] The feature data is assessed for credit risk based on a preset credit scoring card model to obtain a first scoring result, and the fused feature data is assessed for credit risk based on a preset credit scoring model to obtain a second scoring result.

[0161] The fused feature data is obtained by fusing the initial feature data and the feature data; the preset credit scoring card model is obtained based on the feature parameters and the weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data.

[0162] The total score is calculated based on the first score, the second score, and the hyperparameters of the adaptive adjustment; the hyperparameters of the adaptive adjustment are obtained from the evaluation index parameters of the model prediction results of the machine learning model.

[0163] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0164] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0165] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0166] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0167] In the description of this specification, the references to terms such as "an embodiment," "a specific embodiment," "some embodiments," "for example," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0168] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for assessing and processing corporate credit risk, characterized in that, include: Acquire relevant enterprise data, preprocess the relevant enterprise data, and obtain initial feature data; The initial feature data is subjected to feature engineering to obtain feature data that has a significant impact on corporate credit risk assessment; The feature data is assessed for credit risk based on a preset credit scoring card model to obtain a first scoring result, and the fused feature data is assessed for credit risk based on a preset credit scoring model to obtain a second scoring result. The fused feature data is obtained by fusing the initial feature data and the feature data, and the fusion is a concatenation operation; the preset credit scoring card model is obtained based on feature parameters and weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data. The total score is calculated based on the first score, the second score, and the hyperparameters of the adaptive adjustment; the hyperparameters of the adaptive adjustment are obtained from the evaluation index parameters of the model prediction results of the machine learning model. The step of calculating the total score based on the first score result, the second score result, and the adaptive adjustment hyperparameter includes: using the difference between 1 and the square of the adaptive adjustment hyperparameter as the first weight value of the first score result, and using the square of the adaptive adjustment hyperparameter as the second weight value of the second score result; and obtaining the total score based on the first weight value, the first score result, the second weight value, and the second score result.

2. The enterprise credit risk assessment and processing method according to claim 1, characterized in that, The evaluation index parameters of the model prediction results include the area under the receiver operating characteristic curve and the coordinate axis; correspondingly, the hyperparameters of the adaptive modulation are obtained based on the evaluation index parameters of the model prediction results of the machine learning model, including: The area corresponding value is assigned to the hyperparameter of the adaptive adjustment.

3. The enterprise credit risk assessment and processing method according to claim 2, characterized in that, The enterprise credit risk assessment and processing method also includes: Within a preset initial period for credit risk assessment of fused feature data based on a preset credit scoring model, the hyperparameters of the adaptive adjustment are gradually increased from zero, and the hyperparameters of the adaptive adjustment are no longer increased when they reach a preset amplitude value. When the preset initial time period is reached, the step of assigning the area corresponding value to the hyperparameter of the adaptive adjustment is performed.

4. The enterprise credit risk assessment and processing method according to any one of claims 1 to 3, characterized in that, The machine learning model includes at least one or more of the following: decision tree model, optimized distributed gradient boosting library, logistic regression model, and long short-term memory network model.

5. The enterprise credit risk assessment and processing method according to any one of claims 1 to 3, characterized in that, The enterprise credit risk assessment and processing method also includes: Generate a credit analysis report and visualize the feature data and the credit analysis report.

6. The enterprise credit risk assessment and processing method according to claim 5, characterized in that, The enterprise credit risk assessment and processing method also includes: The system responds to the credit loan results obtained by the user based on the feature data and the credit analysis report, and updates the enterprise-related sample data in the training set based on the credit loan results.

7. A corporate credit risk assessment and processing device, characterized in that, include: The acquisition unit is used to acquire enterprise-related data, preprocess the enterprise-related data, and obtain initial feature data; The processing unit is used to perform feature engineering on the initial feature data to obtain feature data that has a significant impact on enterprise credit risk assessment. The evaluation unit is used to perform credit risk assessment on the feature data based on a preset credit scoring card model to obtain a first scoring result, and to perform credit risk assessment on the fused feature data based on a preset credit scoring model to obtain a second scoring result. The fused feature data is obtained by fusing the initial feature data and the feature data, and the fusion is a concatenation operation; the preset credit scoring card model is obtained based on feature parameters and weight parameters corresponding to the feature data; the preset credit scoring model is obtained by training a machine learning model based on enterprise-related sample data. The calculation unit is used to calculate the total score result based on the first score result, the second score result, and the hyperparameters of the adaptive adjustment; the hyperparameters of the adaptive adjustment are obtained based on the evaluation index parameters of the model prediction results of the machine learning model. Specifically, the calculation unit is used to: take the difference between 1 and the square of the adaptively adjusted hyperparameter as the first weight value of the first scoring result, take the square of the adaptively adjusted hyperparameter as the second weight value of the second scoring result, and obtain the total scoring result based on the first weight value, the first scoring result, the second weight value, and the second scoring result.

8. 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, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images