Regional subject management method and device based on credit score

Abstract

The invention discloses a region subject management method and device based on credit scores. The method comprises the following steps: obtaining a first index data sequence of multiple dimensions of a target region subject in a first target time period, and determining a corresponding first feature sequence, each sequence including a plurality of first index data affecting credit assessment; analyzing each feature sequence by using a pre-trained credit risk assessment model comprising a gating cycle unit layer to obtain a first default probability value of each dimension; converting the default probability value into a first credit score through a score card model; determining a weight coefficient of each dimension, and calculating a comprehensive credit score in combination with the credit score; and determining a target credit rating corresponding to the comprehensive score according to a preset credit rating management table, and executing a corresponding management strategy. According to the method and the device, the technical problems of strong subjectivity and lack of objective and comprehensive evaluation of credit evaluation for region subjects in a credit score construction scene are solved.

Description

Technical Field

[0001] The present application relates to the field of information processing technology, and in particular to a regional entity management method and device based on credit scoring. Background Art

[0002] Credit scoring plays a fundamental role in business cooperation, bidding, debt collection and other scenarios. It can help enterprises quantify and evaluate the credit risk of cooperative entities, thereby reducing economic losses caused by credit risk. Unlike general credit assessments for enterprises, the assessment of regional special entities must consider multiple factors. In traditional credit scoring methods, the assessment of regional special entities mainly relies on expert experience, that is, experts with rich domain knowledge manually set the assessment method and scoring threshold based on past experience. Although this method has certain effectiveness, it also has the following problems: expert scoring introduces personal subjective consciousness, which may lead to non-standardization and inconsistency of scoring results. Traditional expert evaluation is difficult to fully utilize massive data, especially for the processing of time series data, so that the scoring results may not be accurate enough and difficult to migrate in different scenarios. For the assessment of regional entities, traditional methods do not fully apply advanced time series neural networks, and it is difficult to capture risk changes in historical information, which affects the scientificity and comprehensiveness of the assessment.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The embodiments of the present application provide a regional entity management method and device based on credit scoring, so as to at least solve the technical problem that the credit assessment of regional entities in the credit scoring construction scenario is highly subjective and lacks objective and comprehensive evaluation.

[0005] According to one aspect of an embodiment of the present application, a regional entity management method based on credit scoring is provided, comprising: obtaining multiple first indicator data sequences under multiple dimensions of a target regional entity within a first target time period, and determining a first feature sequence corresponding to each first indicator data sequence, wherein each first indicator data sequence includes multiple first indicator data that affect the credit evaluation of the target regional entity under one dimension; analyzing each first feature sequence using a pre-trained credit risk assessment model to obtain a first default probability value for each dimension, and determining a first credit score corresponding to the first default probability value for each dimension using a scoring card model, wherein the credit risk assessment model at least includes: a gated recurrent unit layer; determining a weight coefficient for each dimension, and determining a comprehensive credit score of the target regional entity using the weight coefficient for each dimension and the first credit score; determining a target credit rating corresponding to the comprehensive credit score according to a preset credit rating management table, and executing a target management strategy corresponding to the target credit rating for the target regional entity.

[0006] Optionally, the dimensions of the first indicator data series include at least one of the following: macroeconomic conditions dimension, financial soundness dimension, environmental impact dimension, and historical transaction performance dimension; the data of the macroeconomic conditions dimension include at least one of the following: economic scale indicators, economic growth indicators, and investment activity level indicators reflecting the region where the regional entity is located; the data of the financial soundness dimension include at least one of the following: financial ratio indicators, debt level indicators, debt repayment ability indicators, and financial stability indicators reflecting the financial status and debt management of the regional entity; the data of the environmental impact dimension include: various environmental indicators for assessing the credit risk of the regional entity, and various environmental indicators reflect the industry status and regional economic environment of the regional entity in the environment in which it is located; the data of the historical transaction performance dimension include at least one of the following: transaction records or default records between the regional entity and the assessment agency.

[0007] Optionally, multiple first indicator data sequences under multiple dimensions of the target area subject within the first target time period are obtained, and a first feature sequence corresponding to each first indicator data sequence is determined, including: for each first indicator data sequence, all first indicator data in the first indicator data sequence are preprocessed to obtain multiple second indicator data, wherein the data preprocessing method includes at least one of the following: outlier deletion processing, null value interpolation processing, and data standardization processing; the first feature corresponding to each second indicator data is extracted, and multiple first features are screened using a target screening method, wherein the target screening method includes at least one of the following: determining the correlation coefficient between each first feature, and clustering all first features according to the correlation coefficient to obtain multiple clusters, retaining one first feature in each cluster; determining the importance score of each first feature based on a tree model, and removing the first feature whose importance score is less than a preset score threshold, wherein the tree model includes at least one of the following: a random forest tree and a gradient boosting tree; determining the binned variable distribution map corresponding to the multiple first features, and removing the first features that do not meet the preset screening criteria according to the binned variable distribution map; arranging the screened first features in chronological order to obtain a first feature sequence.

[0008] Optionally, the training process of the credit risk assessment model includes: obtaining multiple second indicator data sequences of multiple regional entities in multiple dimensions within a historical time period, and determining the second feature sequence corresponding to each second indicator data sequence, and taking each second feature sequence as a training sample; determining the credit status of each regional entity based on aging analysis or a preset black and white list, and taking the credit status as a sample label of multiple second feature sequences corresponding to the regional entity; constructing an initial assessment model, the initial assessment model at least including: an input layer, a gated recurrent unit layer and an output layer, the gated recurrent unit layer is used to determine the long-term dependency between the input feature sequence data, and update the hidden state at each time step, and the output layer is used to convert the hidden state into a second default probability value; iteratively training the initial assessment model using multiple training samples and sample labels, and taking the model that reaches the preset performance indicators as the credit risk assessment model, wherein, during the training process, the loss function is determined based on the second default probability value, and the model parameters are adjusted based on the loss function combined with the back propagation algorithm.

[0009] Optionally, determining the first credit score corresponding to the first default probability value of each dimension by using the scorecard model includes: converting the first default probability value of each dimension into the first credit score by using a scorecard formula, wherein the scorecard formula is: A i =Score0+B i *ln(θ0), In the formula, Score i represents the first credit score of the i-th dimension, A i With B i is the scoring parameter, p i represents the first default probability value of the i-th dimension, Score0 is the preset benchmark score, θ0 is the preset constant, PDO i is the preset score spread of the i-th dimension, indicating that when the ratio When doubled, the first credit score of the corresponding i-th dimension increases by a certain percentage.

[0010] Optionally, determine the weight coefficient of each dimension, and use the weight coefficient of each dimension and the first credit score to determine the comprehensive credit score of the subject in the target area, including: using a three-level scoring scale to construct a priority judgment matrix, each element in the priority judgment matrix represents the importance of each dimension relative to other dimensions in determining the comprehensive credit score; fuzzifying each element in the priority judgment matrix to obtain a fuzzy consistent judgment matrix; converting the fuzzy consistent judgment matrix into a reciprocal matrix; iteratively normalizing the reciprocal matrix to obtain the weight coefficient corresponding to each dimension; and weighted summing the first credit scores of each dimension according to the weight coefficient of each dimension to obtain a comprehensive credit score.

[0011] Optionally, the comprehensive credit score of the subject in the target area is determined using the weight coefficient of each dimension and the first credit score, including: determining the first comprehensive credit score based on the first credit scores of the macroeconomic status dimension, the financial soundness dimension and the environmental impact dimension and the corresponding weight coefficients; and / or determining the second comprehensive credit score based on the first credit scores of the macroeconomic status dimension, the financial soundness dimension, the environmental impact dimension, and the historical transaction performance dimension and the corresponding weight coefficients.

[0012] Optionally, a target credit rating corresponding to the comprehensive credit score is determined based on a preset credit rating management table, and a target management strategy corresponding to the target credit rating is executed on the target region entity, including: determining the target credit rating and target management strategy corresponding to the comprehensive credit score from the credit rating management table, wherein the credit rating management table stores a mapping relationship between multiple preset credit ratings and credit scores, and a management strategy corresponding to each credit rating, and the management strategy includes at least one of the following: adjusting credit limit, loan interest rate, contract level, transaction priority; executing the target management strategy on the target region entity.

[0013] According to another aspect of an embodiment of the present application, a regional entity management device based on credit scoring is also provided, including: an acquisition module, used to acquire multiple first indicator data sequences of a target regional entity under multiple dimensions within a first target time period, and determine a first feature sequence corresponding to each first indicator data sequence, wherein each first indicator data sequence includes multiple first indicator data that affect the credit evaluation of the target regional entity under one dimension; an analysis module, used to analyze each first feature sequence using a pre-trained credit risk assessment model to obtain a first default probability value for each dimension, and use a scoring card model to determine a first credit score corresponding to the first default probability value for each dimension, wherein the credit risk assessment model at least includes: a gated recurrent unit layer; a scoring module, used to determine a weight coefficient for each dimension, and use the weight coefficient for each dimension and the first credit score to determine a comprehensive credit score of the target regional entity; a management module, used to determine a target credit rating corresponding to the comprehensive credit rating according to a preset credit rating management table, and execute a target management strategy corresponding to the target credit rating on the target regional entity.

[0014] According to another aspect of an embodiment of the present application, a computer program product is also provided, the computer program product comprising: a computer program, wherein when the computer program is executed by a processor, the above-mentioned regional entity management method based on credit scoring is implemented.

[0015] According to another aspect of an embodiment of the present application, an electronic device is also provided, which includes: a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the above-mentioned credit score-based regional entity management method through the computer program.

[0016] In the embodiment of the present application, by collecting and analyzing multi-dimensional data, the indicator data sequence of the subject in the target area is obtained from multiple dimensions and converted into a feature sequence. This feature allows the evaluation model to capture all-round information on the subject's credit status, improving the comprehensiveness and depth of the evaluation. The first feature sequence is analyzed using the credit risk assessment model of the gated recurrent unit layer to obtain the first default probability value under each dimension. The design of the gated recurrent unit layer can effectively handle the long-term dependencies in the sequence data, ensure that the model can take into account historical data and dynamic changes, and improve the accuracy of default prediction and the prediction ability of the model. The first default probability value is converted into the first credit score through the scorecard model. This process enables the risk assessment results to be presented in a standardized and quantitative form, which is convenient for enterprises to understand and apply, while ensuring that the score is inversely proportional to the risk level. By determining the weight coefficient of each dimension and applying it to the weighted sum of the first credit score, the comprehensive credit score of the subject in the target area can be calculated. The weight coefficient reflects the importance of each dimension in the total score. It is determined by scientific methods such as fuzzy hierarchical analysis to ensure the objectivity and scientificity of the scoring results. According to the comprehensive credit score, refer to the preset credit rating management table to determine the target credit rating. Based on the credit rating, the enterprise can implement the corresponding target management strategy. The dynamic adjustment of these strategies helps the enterprise optimize resource allocation and improve transaction efficiency with entities with good credit. At the same time, it effectively controls the transaction risk with entities with higher credit risk, thereby solving the technical problem of strong subjectivity and lack of objective and comprehensive evaluation of credit assessment for regional entities in the credit scoring construction scenario. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 is a flow chart of an optional regional subject management method based on credit scoring according to an embodiment of the present application;

[0019] Figure 2 is a schematic structural diagram of an optional regional subject management device based on credit scoring according to an embodiment of the present application;

[0020] Figure 3 It is a schematic diagram of the structure of an optional electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0022] It should be noted that the terms "first", "second", etc. in the specification, claims and drawings of the present application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0023] In order to better understand the embodiments of the present application, some nouns or terms that appear in the description of the embodiments of the present application are first translated and explained as follows:

[0024] Regional subject: refers to the special subject of a region, that is, an institution or entity with a regional background and special characteristics.

[0025] Assessment agency: refers to an agency that needs to analyze and assess the credit status of regional entities and make corresponding credit or business strategies for regional entities based on the assessment results, such as credit rating companies, banks, investment institutions, etc.

[0026] Example 1

[0027] According to an embodiment of the present application, a regional entity management method based on credit scoring is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0028] Figure 1 is a flow chart of a regional subject management method based on credit scoring provided in an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0029] Step S102, obtaining a plurality of first indicator data sequences of a target region subject in a plurality of dimensions within a first target time period, and determining a first feature sequence corresponding to each first indicator data sequence, wherein each first indicator data sequence includes a plurality of first indicator data affecting the credit assessment of the target region subject in one dimension;

[0030] Step S104, using a pre-trained credit risk assessment model to analyze each first feature sequence to obtain a first default probability value of each dimension, and using a scorecard model to determine a first credit score corresponding to the first default probability value of each dimension, wherein the credit risk assessment model at least includes: a gated recurrent unit layer;

[0031] Step S106, determining a weight coefficient for each dimension, and determining a comprehensive credit score of the subject in the target region using the weight coefficient for each dimension and the first credit score;

[0032] Step S108, determining the target credit rating corresponding to the comprehensive credit score according to the preset credit rating management table, and executing the target management strategy corresponding to the target credit rating for the subject in the target region.

[0033] The following describes the various steps of the regional entity management method based on credit scoring in conjunction with the specific implementation process.

[0034] First, multiple first indicator data sequences under multiple dimensions of the subject in the target area within the first target time period are obtained, and a first feature sequence corresponding to each first indicator data sequence is determined, wherein each first indicator data sequence includes multiple first indicator data affecting the credit assessment of the subject in the target area under one dimension.

[0035] As an optional implementation, the dimension of the first indicator data series includes at least one of the following: a macroeconomic status dimension, a financial soundness dimension, an environmental impact dimension, and a historical transaction performance dimension.

[0036] The data of the macroeconomic status dimension include at least one of the following: economic scale indicators, economic growth indicators, and investment activity level indicators reflecting the region where the regional entity is located; among them, the data of the financial soundness dimension include at least one of the following: financial ratio indicators, debt level indicators, debt repayment capacity indicators, and financial stability indicators reflecting the financial status and debt management of the regional entity; the data of the environmental impact dimension include: various environmental indicators for assessing the credit risk of the regional entity, and various environmental indicators reflect the industry status and regional economic environment of the regional entity in the environment in which it is located; the data of the historical transaction performance dimension include at least one of the following: transaction records or default records between the regional entity and the assessment agency.

[0037] Among them, the macroeconomic situation dimension aims to evaluate the overall economic operation of the region where the regional entity is located. Specifically, it focuses on whether the economic growth in the region is stable, whether there are significant economic fluctuations, and whether the expectations for future development are optimistic. By comprehensively considering multiple economic indicators, it ensures a comprehensive understanding of the regional economy. The financial robustness dimension aims to evaluate the financial management capabilities of the regional entity, including debt levels, liquidity and asset allocation. By analyzing financial statements and related data, the financial health and risk management capabilities of the entity are evaluated to ensure that it has good financial management capabilities. The environmental impact dimension aims to evaluate industry trends and their impact on the operation of the entity, including market competition, technological innovation and changes in industry standards. By analyzing these external factors, the competitiveness and adaptability of the entity in the industry are evaluated to help identify potential risks and opportunities. The historical transaction performance dimension aims to evaluate the historical transaction records of the regional entity, including its past performance and default history. By analyzing historical transaction data, the credit performance and performance of the entity are evaluated to provide a basis for predicting its future credit risk.

[0038] As an optional implementation, obtaining multiple first indicator data sequences of the target area subject in multiple dimensions within the first target time period and determining the first feature sequence corresponding to each first indicator data sequence may be performed in the following manner:

[0039] For each first indicator data sequence, data preprocessing is performed on all first indicator data in the first indicator data sequence to obtain a plurality of second indicator data, wherein the data preprocessing method includes at least one of the following: outlier deletion processing, null value interpolation processing, and data standardization processing;

[0040] Extract the first feature corresponding to each second indicator data, and perform feature screening on the multiple first features using a target screening method, wherein the target screening method includes at least one of the following:

[0041] Determine the correlation coefficient between each first feature, and cluster all first features according to the correlation coefficient to obtain multiple clusters, retaining one first feature in each cluster. The correlation coefficient is a statistic that measures the degree of linear correlation between two variables. If two features are highly correlated, they may provide similar information in the model. Through cluster analysis, you group these highly correlated features and retain one feature in each cluster to reduce redundancy between features and avoid overfitting of the model.

[0042] Determine an importance score for each first feature based on a tree model, and remove first features whose importance score is less than a preset score threshold, wherein the tree model includes at least one of the following: a random forest tree and a gradient boosting tree;

[0043] Determine the binned variable distribution graphs corresponding to the multiple first features, and remove the first features that do not meet the preset screening criteria based on the binned variable distribution graphs. The preset screening criteria may include limiting screening through IV (Information Value) thresholds, monotonicity requirements, sample quantity requirements, etc.

[0044] After screening according to any of the above methods, the screened first features are arranged in chronological order to obtain a first feature sequence.

[0045] After obtaining the first feature sequence, each first feature sequence is analyzed using a pre-trained credit risk assessment model to obtain a first default probability value for each dimension, and a scoring card model is used to determine a first credit score corresponding to the first default probability value for each dimension, wherein the credit risk assessment model includes at least: a gated recurrent unit layer.

[0046] As an optional implementation, the training process of the credit risk assessment model may include the following process:

[0047] S1, obtaining multiple second indicator data sequences of multiple regional entities in multiple dimensions within a historical time period, and determining a second feature sequence corresponding to each second indicator data sequence, and taking each second feature sequence as a training sample;

[0048] S2, determining the credit status of each regional entity based on aging analysis or a preset blacklist and whitelist, and using the credit status as a sample label of multiple second feature sequences corresponding to the regional entity;

[0049] Among them, aging analysis is a common method for assessing credit risk. It mainly focuses on the transaction history of regional entities, especially the repayment of debts. In this process: a period of time (observation period) is set to record the debt behavior of the regional entity, and then observe whether there is any overdue repayment in the subsequent period of time (performance period). Based on the data of the observation period, the overdue rate (the proportion of debts not repaid on time) and the collection rate (the proportion of debts repaid during the performance period) of the regional entity are calculated. According to the overdue rate and the collection rate, the credit status of the regional entity can be determined. For example, if the overdue rate is high and the collection rate is low, then the regional entity may be classified as high risk, otherwise it may be low risk.

[0050] Blacklist and whitelist are another method to determine credit status, which directly marks the credit status of regional entities based on known positive or negative information (i.e. blacklist and whitelist). This usually involves historical data and expert judgment: the blacklist represents a list of regional entities with known poor credit, which may include entities with a history of default, and the whitelist represents a list of regional entities with known good credit, which may include entities that have always repaid on time. Depending on whether the regional entity appears in the blacklist and whitelist, it is directly assigned a credit status of default or non-default. When training the model, the model will learn the mapping relationship from feature sequence to credit status (sample label) so that it can predict the credit status of unknown regions based on new feature data in the future.

[0051] S3, constructing an initial evaluation model, the initial evaluation model at least includes: an input layer, a gated recurrent unit layer and an output layer, the gated recurrent unit layer is used to determine the long-term dependency relationship between the input feature sequence data, update the hidden state at each time step, and the output layer is used to convert the hidden state into a second default probability value;

[0052] Among them, the gated recurrent unit layer is the core part of the model, which is used to capture and utilize long-term dependencies in feature sequence data. In traditional RNN (Recurrent Neural Network), information may disappear or expand as the length of the sequence increases, which is called the "gradient vanishing" and "gradient exploding" problems. The gated recurrent unit layer enables the model to better handle long sequence data by introducing two gate mechanisms: reset gate and update gate. The reset gate determines how to reset the hidden state of the previous moment to decide which information to keep and which information to discard. The update gate controls the fusion ratio of the current moment's information and the previous moment's hidden state, and determines the degree of update of the new hidden state. By updating the hidden state at each time step, the gated recurrent unit layer can effectively store and utilize historical information and capture the patterns and trends that change over time in the feature sequence. These hidden states contain important information about the feature sequence and are the key to connecting the input layer and the output layer. The task of the output layer is to convert the hidden state generated by the gated recurrent unit layer into a predicted output, that is, the second default probability value. This probability value reflects the probability that the model predicts that the regional entity will default at a certain point in the future. The output layer usually uses an activation function to ensure that the output is between 0 and 1, so that it can be interpreted as a probability.

[0053] S4, iteratively training the initial evaluation model using multiple training samples and sample labels, and using the model that reaches the preset performance index as the credit risk assessment model, wherein during the training process, the loss function is determined according to the second default probability value, and the model parameters are adjusted according to the loss function combined with the back propagation algorithm.

[0054] The entire model training process involves adjusting the model parameters to minimize the gap between the predicted second default probability value and the actual credit status (sample label). This process is completed by calculating the loss function (such as binary cross entropy loss) and using the back propagation algorithm to gradually optimize the model parameters and improve the model's prediction accuracy.

[0055] As an optional implementation, determining the first credit score corresponding to the first default probability value of each dimension using the scorecard model can be performed in the following manner: converting the first default probability value of each dimension into the first credit score using a scorecard formula, wherein the scorecard formula is:

[0056]

[0057] A i =Score0+B i *ln(θ0)

[0058]

[0059] In the formula, Score i represents the first credit score of the i-th dimension, A i With B i is the scoring parameter, p i represents the first default probability value of the i-th dimension, Score0 is the preset benchmark score, θ0 is the preset constant, PDO i is the preset score spread of the i-th dimension, indicating that when the ratio When doubling, the corresponding first credit score of the i-th dimension increases by a certain amount. In general, θ0 usually represents the default probability ratio under the baseline condition (for example, when the logarithmic probability is 0).

[0060] After obtaining the first credit score of each dimension, the comprehensive credit score is determined according to the correlation coefficient of each dimension, wherein the weight coefficient of each dimension is determined, and the comprehensive credit score of the subject in the target area is determined by using the weight coefficient of each dimension and the first credit score. The following steps can be taken:

[0061] A priority judgment matrix is ​​constructed using a three-level scale table. Each element in the priority judgment matrix represents the importance of each dimension relative to other dimensions in determining the comprehensive credit score. Each element in the priority judgment matrix is ​​fuzzified to obtain a fuzzy consistent judgment matrix. The fuzzy consistent judgment matrix is ​​converted into a reciprocal matrix. The reciprocal matrix is ​​iteratively normalized to obtain the weight coefficient corresponding to each dimension. The first credit score of each dimension is weighted and summed according to the weight coefficient of each dimension to obtain the comprehensive credit score.

[0062] Among them, the priority judgment matrix is ​​used to quantify the relative importance between different dimensions. The three-level scale is an evaluation tool that divides the importance of dimensions into three levels: 1 (equally important), 2 (one is more important than the other), and 3 (one is much more important than the other). When constructing the priority judgment matrix, the evaluator needs to give each dimension a score of importance relative to other dimensions based on professional knowledge and experience, forming an n×n matrix, where n is the number of dimensions. The fuzzy consistent judgment matrix is ​​an improvement on the priority judgment matrix, which deals with the uncertainty of judgment by introducing fuzziness. In this step, the evaluator may not have an absolute sense of certainty about the importance of each dimension, so fuzzy logic is used to describe the imprecision of judgment, such as defining the importance of each dimension relative to other dimensions as a range rather than an exact value. The reciprocal matrix is ​​a special form of the fuzzy consistent judgment matrix, which maintains the reciprocity of the matrix (that is, the elements in the matrix satisfy ), this transformation ensures the symmetry and consistency of the matrix, making the weight calculation process more reasonable. Through iterative normalization, the weight coefficient of each dimension can be calculated from the reciprocal matrix, which usually involves solving the eigenvector of the matrix and then normalizing it to ensure that the sum of all weight coefficients is 1. The weight coefficient reflects the contribution of each dimension to the comprehensive credit score and is the key to calculating the comprehensive credit score.

[0063] Based on the above dimensions of obtaining indicator data, the comprehensive scores can be divided into two categories:

[0064] The first comprehensive credit score is determined based on the first credit scores in the macroeconomic status dimension, the financial soundness dimension and the environmental impact dimension and the corresponding weight coefficients; or the second comprehensive credit score is determined based on the first credit scores in the macroeconomic status dimension, the financial soundness dimension, the environmental impact dimension and the historical transaction performance dimension and the corresponding weight coefficients.

[0065] The first comprehensive credit score focuses on the performance of regional entities in terms of macroeconomics, financial status and environmental adaptability, aiming to evaluate their credit status from a relatively objective perspective. The second comprehensive credit score includes dimensions related to the assessment agency's own historical transactions, which can help the assessment agency better understand the risks of cooperating with entities in specific regions and adjust the scores based on their own experience, so that the scores are more in line with actual business needs. These two scores can be used in different scenarios. For example, the first comprehensive credit score may be more suitable for preliminary screening, while the second comprehensive credit score provides deeper insights in specific cooperation decisions. Such a scoring design fully considers the comprehensiveness and practicality of information, which helps enterprises or institutions make more reasonable and accurate judgments in credit assessment.

[0066] After obtaining the above-mentioned comprehensive credit score, the target credit rating corresponding to the comprehensive credit score is determined according to the preset credit rating management table, and the target management strategy corresponding to the target credit rating is implemented for the target regional entity. This process can be carried out in the following ways:

[0067] The target credit rating and target management strategy corresponding to the comprehensive credit score are determined from the credit rating management table, wherein the credit rating management table stores a mapping relationship between multiple preset credit ratings and credit scores, and a management strategy corresponding to each credit rating, and the management strategy includes at least one of the following: adjusting credit limit, loan interest rate, contract level, transaction priority; executing the target management strategy on the subject in the target area.

[0068] The credit rating management table is a set of preset mapping rules that defines the association between different credit score intervals and credit ratings. For example, a regional entity with a comprehensive credit score of more than 800 points may be rated AAA, indicating a very low risk of default; while a score below 600 points may be rated B, indicating a higher risk of default. The credit rating of the target regional entity can be determined by comparing its comprehensive credit score with the credit rating management table. The credit rating management table not only contains the mapping from credit score to credit rating, but also lists in detail the management strategy corresponding to each credit rating. Once the credit rating and corresponding management strategy of the target regional entity are determined, the enterprise or institution can take action to adjust the transaction conditions or risk management measures with the regional entity according to the strategy. For example, if a regional entity is rated AAA, then the enterprise may decide to increase its credit limit, reduce the loan interest rate, and give it a higher transaction priority to promote the business relationship between the two parties. Conversely, for regional entities with lower ratings, the enterprise may increase the loan interest rate and reduce the transaction priority to reduce credit risk.

[0069] In the embodiment of the present application, by collecting and analyzing multi-dimensional data, the indicator data sequence of the subject in the target area is obtained from multiple dimensions and converted into a feature sequence. This feature allows the evaluation model to capture all-round information on the subject's credit status, improving the comprehensiveness and depth of the evaluation. The first feature sequence is analyzed using the credit risk assessment model of the gated recurrent unit layer to obtain the first default probability value under each dimension. The design of the gated recurrent unit layer can effectively handle the long-term dependencies in the sequence data, ensure that the model can take into account historical data and dynamic changes, and improve the accuracy of default prediction and the prediction ability of the model. The first default probability value is converted into the first credit score through the scorecard model. This process enables the risk assessment results to be presented in a standardized and quantitative form, which is convenient for enterprises to understand and apply, while ensuring that the score is inversely proportional to the risk level. By determining the weight coefficient of each dimension and applying it to the weighted sum of the first credit score, the comprehensive credit score of the subject in the target area can be calculated. The weight coefficient reflects the importance of each dimension in the total score. It is determined by scientific methods such as fuzzy hierarchical analysis to ensure the objectivity and scientificity of the scoring results. According to the comprehensive credit score, refer to the preset credit rating management table to determine the target credit rating. Based on the credit rating, the enterprise can implement the corresponding target management strategy. The dynamic adjustment of these strategies helps the enterprise optimize resource allocation and improve transaction efficiency with entities with good credit. At the same time, it effectively controls the transaction risk with entities with higher credit risk, thereby solving the technical problem of strong subjectivity and lack of objective and comprehensive evaluation of credit assessment for regional entities in the credit scoring construction scenario.

[0070] Example 2

[0071] According to an embodiment of the present application, a regional subject management device based on credit score is also provided for implementing the regional subject management method based on credit score in Example 1, such as Figure 2 As shown, the regional subject management device based on credit scoring at least includes: an acquisition module 21, an analysis module 22, a scoring module 23 and a management module 24, wherein:

[0072] An acquisition module 21 is used to acquire a plurality of first indicator data sequences of a target region subject in a plurality of dimensions within a first target time period, and determine a first feature sequence corresponding to each first indicator data sequence, wherein each first indicator data sequence includes a plurality of first indicator data affecting the credit assessment of the target region subject in one dimension;

[0073] The analysis module 22 is used to analyze each first feature sequence using a pre-trained credit risk assessment model to obtain a first default probability value of each dimension, and determine a first credit score corresponding to the first default probability value of each dimension using a scorecard model, wherein the credit risk assessment model at least includes: a gated recurrent unit layer;

[0074] A scoring module 23, used to determine a weight coefficient for each dimension, and determine a comprehensive credit score of the subject in the target region using the weight coefficient for each dimension and the first credit score;

[0075] The management module 24 is used to determine the target credit rating corresponding to the comprehensive credit score according to the preset credit rating management table, and to execute the target management strategy corresponding to the target credit rating for the subject in the target area.

[0076] The functions of each module of the regional entity management device based on credit scoring are explained below in conjunction with a specific implementation process.

[0077] First, the acquisition module acquires multiple first indicator data sequences under multiple dimensions of the target area subject within the first target time period, and determines a first feature sequence corresponding to each first indicator data sequence, wherein each first indicator data sequence includes multiple first indicator data affecting the credit assessment of the target area subject under one dimension.

[0078] As an optional implementation, the dimension of the first indicator data series includes at least one of the following: a macroeconomic status dimension, a financial soundness dimension, an environmental impact dimension, and a historical transaction performance dimension.

[0079] The data of the macroeconomic status dimension include at least one of the following: economic scale indicators, economic growth indicators, and investment activity level indicators reflecting the region where the regional entity is located; among them, the data of the financial soundness dimension include at least one of the following: financial ratio indicators, debt level indicators, debt repayment capacity indicators, and financial stability indicators reflecting the financial status and debt management of the regional entity; the data of the environmental impact dimension include: various environmental indicators for assessing the credit risk of the regional entity, and various environmental indicators reflect the industry status and regional economic environment of the regional entity in the environment in which it is located; the data of the historical transaction performance dimension include at least one of the following: transaction records or default records between the regional entity and the assessment agency.

[0080] The macroeconomic situation dimension aims to assess the overall economic operation of the region where the regional entity is located. Specifically, it focuses on whether the economic growth in the region is stable, whether there are significant economic fluctuations, and whether the expectations for future development are optimistic. By comprehensively considering multiple economic indicators, it ensures a comprehensive understanding of the regional economy. The financial robustness dimension aims to assess the financial management capabilities of the regional entity, including debt levels, liquidity and asset allocation. By analyzing financial statements and related data, the financial health and risk management capabilities of the entity are assessed to ensure that it has good financial management capabilities. The environmental impact dimension aims to assess industry trends and their impact on the operation of the entity, including market competition, technological innovation and changes in industry standards. By analyzing these external factors, the competitiveness and adaptability of the entity in the industry are assessed to help identify potential risks and opportunities. The historical transaction performance dimension aims to assess the historical transaction records of the regional entity, including its past performance and default history. By analyzing historical transaction data, the credit performance and performance of the entity are assessed to provide a basis for predicting its future credit risk.

[0081] As an optional implementation, obtaining multiple first indicator data sequences of the target area subject in multiple dimensions within the first target time period and determining the first feature sequence corresponding to each first indicator data sequence may be performed in the following manner:

[0082] For each first indicator data sequence, data preprocessing is performed on all first indicator data in the first indicator data sequence to obtain a plurality of second indicator data, wherein the data preprocessing method includes at least one of the following: outlier deletion processing, null value interpolation processing, and data standardization processing;

[0083] Extract the first feature corresponding to each second indicator data, and perform feature screening on the multiple first features using a target screening method, wherein the target screening method includes at least one of the following:

[0084] Determine the correlation coefficient between each first feature, and cluster all first features according to the correlation coefficient to obtain multiple clusters, retaining one first feature in each cluster. The correlation coefficient is a statistic that measures the degree of linear correlation between two variables. If two features are highly correlated, they may provide similar information in the model. Through cluster analysis, you group these highly correlated features and retain one feature in each cluster to reduce redundancy between features and avoid overfitting of the model.

[0085] Determine an importance score for each first feature based on a tree model, and remove first features whose importance score is less than a preset score threshold, wherein the tree model includes at least one of the following: a random forest tree, a gradient boosting tree;

[0086] Determine the binned variable distribution graphs corresponding to the multiple first features, and remove the first features that do not meet the preset screening criteria based on the binned variable distribution graphs. The preset screening criteria may include limited screening through IV thresholds, monotonicity requirements, sample quantity requirements, etc.

[0087] After screening according to any of the above methods, the screened first features are arranged in chronological order to obtain a first feature sequence.

[0088] After obtaining the first feature sequence, the analysis module uses a pre-trained credit risk assessment model to analyze each first feature sequence to obtain a first default probability value for each dimension, and uses a scoring card model to determine a first credit score corresponding to the first default probability value for each dimension, wherein the credit risk assessment model includes at least: a gated recurrent unit layer.

[0089] As an optional implementation, the training process of the credit risk assessment model may include the following process:

[0090] S1, obtaining multiple second indicator data sequences of multiple regional entities in multiple dimensions within a historical time period, and determining a second feature sequence corresponding to each second indicator data sequence, and taking each second feature sequence as a training sample;

[0091] S2, determining the credit status of each regional entity based on aging analysis or a preset blacklist and whitelist, and using the credit status as a sample label of multiple second feature sequences corresponding to the regional entity;

[0092] S3, constructing an initial evaluation model, the initial evaluation model at least includes: an input layer, a gated recurrent unit layer and an output layer, the gated recurrent unit layer is used to determine the long-term dependency relationship between the input feature sequence data, update the hidden state at each time step, and the output layer is used to convert the hidden state into a second default probability value;

[0093] S4, iteratively training the initial evaluation model using multiple training samples and sample labels, and using the model that reaches the preset performance index as the credit risk assessment model, wherein during the training process, the loss function is determined according to the second default probability value, and the model parameters are adjusted according to the loss function combined with the back propagation algorithm.

[0094] The entire model training process involves adjusting the model parameters to minimize the gap between the predicted second default probability value and the actual credit status (sample label). This process is completed by calculating the loss function (such as binary cross entropy loss) and using the back propagation algorithm to gradually optimize the model parameters and improve the model's prediction accuracy.

[0095] As an optional implementation, determining the first credit score corresponding to the first default probability value of each dimension using the scorecard model can be performed in the following manner: converting the first default probability value of each dimension into the first credit score using a scorecard formula, wherein the scorecard formula is:

[0096]

[0097] A i =Score0+B i *ln(θ0)

[0098]

[0099] In the formula, Score i represents the first credit score of the i-th dimension, A i With B i is the scoring parameter, p i represents the first default probability value of the i-th dimension, Score0 is the preset benchmark score, θ0 is the preset constant, PDO i is the preset score spread of the i-th dimension, indicating that when the ratio When doubled, the first credit score of the corresponding i-th dimension increases by a certain percentage.

[0100] After obtaining the first credit score of each dimension, the comprehensive credit score is determined according to the correlation coefficient of each dimension, wherein the scoring module determines the weight coefficient of each dimension, and the process of determining the comprehensive credit score of the subject in the target region by using the weight coefficient of each dimension and the first credit score can take the following steps:

[0101] A priority judgment matrix is ​​constructed using a three-level scale table. Each element in the priority judgment matrix represents the importance of each dimension relative to other dimensions in determining the comprehensive credit score. Each element in the priority judgment matrix is ​​fuzzified to obtain a fuzzy consistent judgment matrix. The fuzzy consistent judgment matrix is ​​converted into a reciprocal matrix. The reciprocal matrix is ​​iteratively normalized to obtain the weight coefficient corresponding to each dimension. The first credit score of each dimension is weighted and summed according to the weight coefficient of each dimension to obtain the comprehensive credit score.

[0102] Based on the above dimensions of obtaining indicator data, the comprehensive scores can be divided into two categories:

[0103] The first comprehensive credit score is determined based on the first credit scores in the macroeconomic status dimension, the financial soundness dimension and the environmental impact dimension and the corresponding weight coefficients; or the second comprehensive credit score is determined based on the first credit scores in the macroeconomic status dimension, the financial soundness dimension, the environmental impact dimension and the historical transaction performance dimension and the corresponding weight coefficients.

[0104] After obtaining the above-mentioned comprehensive credit score, the management module determines the target credit rating corresponding to the comprehensive credit score according to the preset credit rating management table, and implements the target management strategy corresponding to the target credit rating for the target regional entity. This process can be carried out in the following ways:

[0105] The target credit rating and target management strategy corresponding to the comprehensive credit score are determined from the credit rating management table, wherein the credit rating management table stores a mapping relationship between multiple preset credit ratings and credit scores, and a management strategy corresponding to each credit rating, and the management strategy includes at least one of the following: adjusting credit limit, loan interest rate, contract level, transaction priority; executing the target management strategy on the subject in the target area.

[0106] It should be noted that each module in the regional entity management device based on credit scoring in the embodiment of the present application corresponds one by one to each implementation step of the regional entity management method based on credit scoring in Example 1. Since a detailed description has been given in Example 1, some details not reflected in this embodiment can be referred to Example 1 and will not be repeated here.

[0107] Example 3

[0108] According to an embodiment of the present application, a computer program product is also provided, which includes a computer program, wherein when the computer program is executed by a processor, the regional entity management method based on credit scoring in Example 1 is implemented.

[0109] According to an embodiment of the present application, a non-volatile storage medium is also provided, which includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the regional entity management method based on credit scoring in Example 1 by running the computer program.

[0110] According to an embodiment of the present application, a processor is also provided, which is used to run a computer program, wherein the computer program executes the regional entity management method based on credit scoring in Example 1 when running.

[0111] According to an embodiment of the present application, an electronic device is also provided, which includes: a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the regional entity management method based on credit scoring in Example 1 through the computer program.

[0112] Specifically, when the computer program is running, the following steps are executed: obtaining multiple first indicator data sequences under multiple dimensions of the target area subject within the first target time period, and determining a first feature sequence corresponding to each first indicator data sequence, wherein each first indicator data sequence includes multiple first indicator data that affect the credit evaluation of the target area subject under one dimension; using a pre-trained credit risk assessment model to analyze each first feature sequence to obtain a first default probability value for each dimension, and using a scoring card model to determine a first credit score corresponding to the first default probability value for each dimension, wherein the credit risk assessment model at least includes: a gated recurrent unit layer; determining a weight coefficient for each dimension, and determining a comprehensive credit score of the target area subject using the weight coefficient for each dimension and the first credit score; determining a target credit rating corresponding to the comprehensive credit rating based on a preset credit rating management table, and executing a target management strategy corresponding to the target credit rating on the target area subject.

[0113] As an optional implementation, the electronic device may be in the form of a mobile terminal, a computer terminal or a similar computing device. Figure 3 FIG. 1 shows a hardware structure block diagram of an electronic device for implementing a regional subject management method based on credit scoring. Figure 3 As shown, the electronic device 30 may include one or more (302a, 302b, ..., 302n are used to illustrate) processors 302 (the processor 302 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 304 for storing data, and a transmission device 306 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 3 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 3 More or fewer components as shown, or with Figure 3 Different configurations shown.

[0114] It should be noted that the one or more processors 302 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the electronic device 30. As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0115] The memory 304 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the regional subject management method based on credit scoring in the embodiment of the present application. The processor 302 executes various functional applications and data processing by running the software programs and modules stored in the memory 304, that is, to implement the vulnerability detection method of the above-mentioned application. The memory 304 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 304 may further include a memory remotely arranged relative to the processor 302, and these remote memories may be connected to the electronic device 30 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 a combination thereof.

[0116] The transmission device 306 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 electronic device 30. In one example, the transmission device 306 includes a network adapter (Network Interface Controller, 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 306 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0117] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the electronic device 30 .

[0118] The serial numbers of the above embodiments are only for description and do not represent the advantages or disadvantages of the embodiments.

[0119] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0120] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0121] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0122] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0123] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0124] The above are only preferred implementations of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A regional entity management method based on credit scoring, characterized in that: include: Acquire multiple first indicator data sequences of the subject in the target region under multiple dimensions within the first target time period, and determine a first feature sequence corresponding to each of the first indicator data sequences, wherein each of the first indicator data sequences includes multiple first indicator data in one dimension that affect the credit assessment of the subject in the target region; Analyze each of the first feature sequences using a pre-trained credit risk assessment model to obtain a first default probability value for each dimension, and determine a first credit score corresponding to the first default probability value for each dimension using a scorecard model, wherein the credit risk assessment model includes at least: a gated recurrent unit layer; Determining a weight coefficient for each dimension, and determining a comprehensive credit score for the subject in the target region using the weight coefficient for each dimension and the first credit score; The target credit rating corresponding to the comprehensive credit score is determined according to a preset credit rating management table, and the target management strategy corresponding to the target credit rating is executed on the target regional entity.

2. The method according to claim 1, characterized in that: The dimension of the first indicator data series includes at least one of the following: a macroeconomic status dimension, a financial robustness dimension, an environmental impact dimension, and a historical transaction performance dimension; The data of the macroeconomic situation dimension include at least one of the following: economic scale indicators, economic growth indicators, and investment activity level indicators reflecting the region where the regional entity is located; The data of the financial robustness dimension include at least one of the following: financial ratio indicators reflecting the financial status and debt management of regional entities, debt level indicators, debt repayment capacity indicators, and financial stability indicators; The data of the environmental impact dimension include: various environmental indicators for assessing the credit risk of regional entities, and the various environmental indicators reflect the industry status and regional economic environment of regional entities in their environment; The data of the historical transaction performance dimension includes at least one of the following: transaction records or default records between the regional entity and the assessment agency.

3. The method according to claim 1, characterized in that Acquiring multiple first indicator data sequences of the target area subject in multiple dimensions within the first target time period, and determining a first feature sequence corresponding to each of the first indicator data sequences, including: For each of the first indicator data sequences, data preprocessing is performed on all the first indicator data in the first indicator data sequence to obtain a plurality of second indicator data, wherein the data preprocessing method includes at least one of the following: outlier deletion processing, null value interpolation processing, and data standardization processing; Extract the first feature corresponding to each of the second indicator data, and perform feature screening on multiple first features using a target screening method, wherein the target screening method includes at least one of the following: determine the correlation coefficient between each first feature, and cluster all the first features according to the correlation coefficient to obtain multiple clusters, and retain one first feature in each cluster; determine the importance score of each first feature based on a tree model, and remove the first feature whose importance score is less than a preset score threshold, and the tree model includes at least one of the following: a random forest tree and a gradient boosting tree; determine the binned variable distribution map corresponding to multiple first features, and remove the first features that do not meet the preset screening criteria according to the binned variable distribution map; The first features obtained by screening are arranged in chronological order to obtain the first feature sequence.

4. The method according to claim 1, characterized in that: The training process of the credit risk assessment model includes: Acquire multiple second indicator data sequences of multiple regional entities in multiple dimensions within a historical time period, determine a second feature sequence corresponding to each of the second indicator data sequences, and use each of the second feature sequences as a training sample; Determine the credit status of each regional entity based on aging analysis or a preset blacklist and whitelist, and use the credit status as a sample label of a plurality of second feature sequences corresponding to the regional entity; Constructing an initial assessment model, the initial assessment model at least comprising: an input layer, a gated recurrent unit layer and an output layer, the gated recurrent unit layer is used to determine the long-term dependency relationship between the input feature sequence data, and update the hidden state at each time step, and the output layer is used to convert the hidden state into a second default probability value; The initial assessment model is iteratively trained using a plurality of the training samples and sample labels, and a model that reaches a preset performance indicator is used as the credit risk assessment model, wherein during the training process, a loss function is determined based on the second default probability value, and model parameters are adjusted based on the loss function combined with a back propagation algorithm.

5. The method according to claim 1, characterized in that Determining a first credit score corresponding to the first default probability value of each dimension by using a scorecard model includes: The first default probability value of each dimension is converted into the first credit score using a scorecard formula, wherein the scorecard formula is: A i =Score0+B i *ln(θ0) In the formula, Score i represents the first credit score of the i-th dimension, A i With B i is the scoring parameter, p i represents the first default probability value of the i-th dimension, Score0 is the preset benchmark score, θ0 is the preset constant, PDO i is the preset score spread of the i-th dimension, indicating that when the ratio When doubled, the first credit score of the corresponding i-th dimension increases by a certain percentage.

6. The method according to claim 1, characterized in that Determining a weight coefficient for each dimension, and determining a comprehensive credit score of the subject in the target region using the weight coefficient for each dimension and the first credit score, including: A priority judgment matrix is ​​constructed using a three-level scoring scale, wherein each element in the priority judgment matrix represents the importance of each dimension relative to other dimensions in determining the comprehensive credit score; Performing fuzzy processing on each element in the priority judgment matrix to obtain a fuzzy consistent judgment matrix; Converting the fuzzy consistent judgment matrix into a reciprocal matrix; Iteratively normalize the reciprocal matrix to obtain a weight coefficient corresponding to each dimension; The first credit scores of each dimension are weighted and summed according to the weight coefficient of each dimension to obtain the comprehensive credit score.

7. The method according to claim 2, characterized in that Determining the comprehensive credit score of the subject in the target region by using the weight coefficient of each dimension and the first credit score includes: Determining a first comprehensive credit score based on the first credit scores of the macroeconomic status dimension, the financial robustness dimension and the environmental impact dimension and corresponding weight coefficients; and / or, A second comprehensive credit score is determined based on the first credit scores in the macroeconomic status dimension, financial soundness dimension, environmental impact dimension, and historical transaction performance dimension and corresponding weight coefficients.

8. The method according to claim 1, characterized in that Determine the target credit rating corresponding to the comprehensive credit score according to the preset credit rating management table, and implement the target management strategy corresponding to the target credit rating for the target regional entity, including: Determining a target credit rating and a target management strategy corresponding to the comprehensive credit score from the credit rating management table, wherein the credit rating management table stores a mapping relationship between a plurality of preset credit ratings and credit scores, and a management strategy corresponding to each credit rating, wherein the management strategy includes at least one of the following: adjusting a credit limit, a loan interest rate, a contract grade, and a transaction priority; The target management strategy is executed on the target area subject.

9. A regional entity management device based on credit scoring, characterized in that: include: The acquisition module is used to acquire a plurality of first indicator data sequences under a plurality of dimensions of the subject in the target region within a first target time period, and determine a first feature sequence corresponding to each of the first indicator data sequences, wherein each of the first indicator data sequences includes a plurality of first indicator data under a dimension that affect the credit assessment of the subject in the target region; The analysis module is used to analyze each of the first feature sequences using a pre-trained credit risk assessment model to obtain a first default probability value for each dimension, and to determine a first credit score corresponding to the first default probability value for each dimension using a scorecard model, wherein the credit risk assessment model at least includes: a gated recurrent unit layer; The scoring module is used to determine a weight coefficient for each dimension, and determine a comprehensive credit score for the subject in the target region using the weight coefficient for each dimension and the first credit score; The management module is used to determine the target credit rating corresponding to the comprehensive credit score according to a preset credit rating management table, and to execute the target management strategy corresponding to the target credit rating for the subject in the target area.

10. An electronic device, characterized in that: include: A memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the regional entity management method based on credit scoring according to any one of claims 1 to 8 through the computer program.

Images