Credit data acquisition method and system based on deep learning, and computer equipment

Abstract

The invention relates to the technical field of data acquisition, in particular to a credit data acquisition method and system based on deep learning and computer equipment. According to the method, cross validation is carried out by integrating the supply chain data, the financial statement and the credit report data, the credit condition of an enterprise can be comprehensively and stereoscopically known, the comprehensiveness, the accuracy and the consistency of the data are improved, and multi-dimensional data such as financial features, credit features and supply chain features are introduced, so that the enterprise credit condition can be comprehensively and stereoscopically known. The method can make up for the missing or deviation possibly existing in a single data source, avoids the misjudgment possibly caused by the single data source, and can more accurately capture the real credit condition of an enterprise through combining a plurality of indexes such as debt paying capability, financial health, credit risk and the like. By inputting the debt paying ability index, the financial health index and the credit risk index into the credit risk prediction model, the model can obtain more input data by using the multi-dimensional and comprehensive information, so that the prediction ability and robustness of the model are improved.

Description

Technical Field

[0001] The present invention relates to the field of data collection technology, and in particular to a credit data collection method, system and computer device based on deep learning. Background Art

[0002] Corporate credit refers to the credit granted by one corporate entity to another corporate entity. Its essence is the monetary loan from the seller to the buyer. It includes the credit sales made by manufacturing companies to corporate clients in credit management, that is, product credit sales. In the process of product credit sales, the credit grantors are usually material suppliers, product manufacturers and wholesalers, while the buyers are the beneficiaries of product credit sales. They are various corporate customers or agents. The buyers obtain the credit granted by the seller in the name of their own companies. Corporate credit also involves the credit of commercial banks, finance companies and other financial institutions to enterprises, as well as the credit generated by trade methods other than sight remittance payment and advance payment.

[0003] Therefore, the collection and evaluation of corporate credit data is very important for the subsequent development of the company. However, traditional corporate credit evaluation may rely on a single data source, such as relying solely on a certain type of data in financial statements or credit reports. This may lead to incomplete data missing, and the credit risk and other indicators obtained through these missing data are not accurate enough. Summary of the invention

[0004] The main purpose of the present invention is to provide a credit data collection method based on deep learning, aiming to solve the technical problems in the prior art.

[0005] The present invention proposes a credit data collection method based on deep learning, comprising:

[0006] Acquiring historical data features of enterprise credit data, wherein the historical data features include historical supply chain data sources, historical enterprise financial statements, and historical enterprise credit reports;

[0007] Cross-validating the historical supply chain data source, historical enterprise financial statements, and historical enterprise credit reports to obtain consistent supply chain data, consistent financial data, and consistent credit data;

[0008] Acquire supply chain data features of the consistent supply chain data, and perform dimensionality reduction processing on the supply chain data features to obtain target supply chain features;

[0009] Acquiring financial data features of the consistent financial data, and performing dimensionality reduction processing on the financial data features to obtain target financial features;

[0010] Acquiring credit data features of the consistent credit data, and performing dimensionality reduction processing on the credit data features to obtain target credit features;

[0011] Calculating a solvency index based on the target financial characteristics and target credit characteristics;

[0012] Calculating a financial health index based on the target supply chain characteristics and the target financial characteristics;

[0013] A credit risk index is obtained according to the target supply chain characteristics and the target credit characteristics, and the debt repayment capacity index, the financial health index and the credit risk index are input into a credit risk prediction model to obtain a credit assessment value.

[0014] Preferably, the step of cross-validating the historical supply chain data source, historical enterprise financial statements and historical enterprise credit reports to obtain consistent supply chain data, consistent financial data and consistent credit data includes:

[0015] Respectively obtain all supply chain data of the historical supply chain data source, all financial data of historical enterprise financial statements, and all credit data of historical enterprise credit reports;

[0016] Extracting data identical to all financial data and all credit data from all supply chain data to obtain consistent supply chain data;

[0017] Extracting the same data as those in all supply chain data and all credit data from all the financial data to obtain consistent financial data;

[0018] The data identical to those in all the financial data and all the supply chain data are extracted from all the credit data to obtain consistent credit data.

[0019] Preferably, the step of performing dimensionality reduction processing on the supply chain data features to obtain target supply chain features comprises:

[0020] Performing data standardization processing on the supply chain data characteristics to obtain standard supply chain characteristics;

[0021] Performing mean filling processing on the standard supply chain characteristics to obtain complete supply chain characteristics;

[0022] Obtaining a supply chain covariance matrix according to the complete supply chain characteristics, and obtaining all supply chain eigenvalues ​​and supply chain eigenvectors of the supply chain covariance matrix;

[0023] Mark the supply chain feature vectors corresponding to multiple supply chain feature values ​​greater than the supply chain threshold as the principal component feature vectors;

[0024] Constructing a projection matrix according to a plurality of the principal component eigenvectors;

[0025] The supply chain covariance matrix is ​​linearly transformed according to the projection matrix and projected onto the principal component space to obtain the target supply chain characteristics.

[0026] Preferably, the step of calculating the debt repayment ability index according to the target financial characteristics and the target credit characteristics comprises:

[0027] Obtaining debt repayment indicators and profitability indicators based on the target financial characteristics, and obtaining current ratio, quick ratio and interest coverage index based on the debt repayment indicators;

[0028] Obtain the debt coverage ratio based on the current ratio, quick ratio and interest coverage index;

[0029] Obtaining a net profit margin and a return on net assets based on the profitability indicator, and obtaining a profitability ratio based on the net profit margin and the return on net assets;

[0030] Obtaining a credit utilization rate and a debt compliance rate based on the target credit characteristics;

[0031] The debt repayment ability index is calculated based on the credit utilization ratio, debt compliance ratio, debt coverage ratio and profitability ratio, wherein the calculation formula is:

[0032]

[0033] Among them, C(ZN) represents the debt repayment ability index, X(LV) represents the credit utilization rate, Z(YL) represents the debt fulfillment rate, Z(FB) represents the debt coverage ratio, and Y(BL) represents the profitability ratio.

[0034] Preferably, the step of calculating the financial health index according to the target supply chain characteristics and the target financial characteristics comprises:

[0035] Acquire inventory management information and supplier management information according to the target supply chain characteristics, and acquire inventory turnover rate and inventory turnover cycle according to the inventory management information;

[0036] Obtaining accounts payable to accounts receivable ratios and purchases to sales ratios based on the supplier management information;

[0037] Obtaining supply chain management efficiency based on the inventory turnover ratio, inventory turnover cycle, accounts payable to accounts receivable ratio and purchase to sales ratio;

[0038] Acquire operation efficiency information and cash flow information according to the target financial characteristics, and acquire inventory turnover rate and accounts receivable turnover rate according to the operation efficiency information;

[0039] Obtaining a cash flow ratio and current liabilities based on the cash flow information, and obtaining operating cash flow based on the cash flow ratio and current liabilities;

[0040] Obtain the debt ratio based on the current liabilities, and obtain the working capital based on the operating cash flow and current liabilities;

[0041] The financial liquidity value is calculated based on the debt ratio, operating cash flow, working capital, inventory turnover rate and accounts receivable turnover rate, where the calculation formula is:

[0042]

[0043] Among them, C(LD) represents financial liquidity value, J(XJ) represents operating cash flow, Y(YZ) represents working capital, C(ZL) represents inventory turnover rate, Y(SL) represents accounts receivable turnover rate, and F(BL) represents debt ratio;

[0044] A financial health index is obtained according to the financial liquidity value and the supply chain management efficiency.

[0045] Preferably, the step of obtaining a credit risk index according to the target supply chain characteristics and the target credit characteristics comprises:

[0046] Acquire supply chain performance data according to the target supply chain characteristics, and acquire supply chain dependence and supply chain stability according to the supply chain performance data;

[0047] Obtaining a delivery on-time rate and a delivery quality rate according to the supply chain stability, and obtaining a timeliness coefficient according to the delivery on-time rate and the supply chain dependence;

[0048] Obtaining a quality compliance coefficient based on the supply chain dependency and the delivery quality rate;

[0049] Obtaining a credit score and a default risk score based on the target credit characteristics;

[0050] A credit risk index is obtained based on the timeliness coefficient, compliance coefficient, credit score and default risk score.

[0051] The present application also provides a credit data collection system based on deep learning, comprising:

[0052] A first acquisition module is used to acquire historical data features of enterprise credit data, wherein the historical data features include historical supply chain data sources, historical enterprise financial statements and historical enterprise credit reports;

[0053] A verification module, used to cross-verify the historical supply chain data source, historical enterprise financial statements and historical enterprise credit reports to obtain consistent supply chain data, consistent financial data and consistent credit data;

[0054] A first dimensionality reduction module is used to obtain supply chain data features of the consistent supply chain data, and perform dimensionality reduction processing on the supply chain data features to obtain target supply chain features;

[0055] A second dimension reduction module is used to obtain the financial data features of the consistent financial data and perform dimension reduction processing on the financial data features to obtain target financial features;

[0056] A third dimension reduction module is used to obtain the credit data features of the consistent credit data, and perform dimension reduction processing on the credit data features to obtain target credit features;

[0057] A first calculation module, used to calculate a debt repayment ability index according to the target financial characteristics and target credit characteristics;

[0058] A second calculation module, used to calculate a financial health index according to the target supply chain characteristics and the target financial characteristics;

[0059] The second acquisition module is used to obtain a credit risk index based on the target supply chain characteristics and target credit characteristics, and input the debt repayment ability index, financial health index and credit risk index into a credit risk prediction model to obtain a credit assessment value.

[0060] Preferably, the second acquisition module includes:

[0061] A first acquisition unit is used to acquire supply chain performance data according to the target supply chain characteristics, and acquire supply chain dependence and supply chain stability according to the supply chain performance data;

[0062] A second acquisition unit is used to acquire a delivery on-time rate and a delivery quality rate according to the supply chain stability, and to acquire a timeliness coefficient according to the delivery on-time rate and the supply chain dependence;

[0063] A third acquisition unit is used to acquire a quality compliance coefficient according to the supply chain dependence and the delivery quality rate;

[0064] A fourth acquisition unit, configured to acquire a credit score and a default risk score according to the target credit feature;

[0065] The fifth acquisition unit is used to acquire a credit risk index according to the timeliness coefficient, the compliance coefficient, the credit score and the default risk score.

[0066] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned credit data collection method based on deep learning when executing the computer program.

[0067] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned credit data collection method based on deep learning.

[0068] The beneficial effects of the present invention are as follows: the present invention can comprehensively and three-dimensionally understand the credit status of an enterprise by cross-validating comprehensive supply chain data, financial statements and credit report data, thereby improving the comprehensiveness, accuracy and consistency of the data; through dimensionality reduction processing, the most representative and effective information can be extracted from a large number of features; the target supply chain features, target financial features and target credit features obtained after dimensionality reduction can better capture the core credit risk information of the enterprise, thereby improving the efficiency and accuracy of the evaluation model; by introducing multi-dimensional data such as financial features, credit features and supply chain features, it can make up for the possible omissions or deviations in a single data source, thereby more comprehensively reflecting the comprehensive status of the enterprise and avoiding the misjudgment that may be caused by a single data source; combining multiple indicators such as debt repayment ability, financial health, credit risk, etc. can more accurately capture the true credit status of the enterprise; by inputting the debt repayment ability index, financial health index and credit risk index into the credit risk prediction model, using these multi-dimensional and comprehensive information, the model can obtain more input data, thereby improving the prediction ability and robustness of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 The figure is a schematic diagram of a method flow according to an embodiment of the present invention.

[0070] Figure 2 FIG. 1 is a schematic diagram of a device structure according to an embodiment of the present invention.

[0071] Figure 3 A schematic diagram of the internal structure of a computer device according to an embodiment of the present application.

[0072] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0073] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0074] like Figure 1-Figure 3 As shown, the present application provides a credit data collection method based on deep learning, comprising:

[0075] S1. Obtaining historical data features of enterprise credit data, wherein the historical data features include historical supply chain data sources, historical enterprise financial statements, and historical enterprise credit reports;

[0076] S2. Cross-validate the historical supply chain data source, historical enterprise financial statements, and historical enterprise credit reports to obtain consistent supply chain data, consistent financial data, and consistent credit data;

[0077] S3, obtaining supply chain data features of the consistent supply chain data, and performing dimensionality reduction processing on the supply chain data features to obtain target supply chain features;

[0078] S4, obtaining financial data features of the consistent financial data, and performing dimensionality reduction processing on the financial data features to obtain target financial features;

[0079] S5. Acquire the credit data features of the consistent credit data, and perform dimensionality reduction processing on the credit data features to obtain target credit features;

[0080] S6. Calculating a debt-paying capacity index according to the target financial characteristics and target credit characteristics;

[0081] S7. Calculating a financial health index according to the target supply chain characteristics and target financial characteristics;

[0082] S8. Obtain a credit risk index based on the target supply chain characteristics and the target credit characteristics, and input the debt repayment capacity index, financial health index and credit risk index into a credit risk prediction model to obtain a credit assessment value.

[0083] As described in the above steps S1-S8, corporate credit generally refers to the credit granted by one corporate entity to another corporate entity. Corporate credit also involves the credit provided to enterprises by commercial banks, finance companies, and other financial institutions, as well as the credit generated by trade methods other than spot remittance payment and prepayment. However, traditional corporate credit assessment may rely on a single data source, such as relying solely on a certain type of data in financial statements or credit reports, which may lead to incomplete data missing, and thus the results of indicators such as credit risk obtained through these missing data are not accurate enough. The present invention obtains consistent supply chain data, consistent financial data and consistent credit data by cross-validating historical supply chain data sources, historical enterprise financial statements and historical enterprise credit reports of enterprise credit data. By cross-validating comprehensive supply chain data, financial statements and credit report data, the credit status of the enterprise can be fully and three-dimensionally understood. Different data sources can complement and cross-verify each other, thereby improving the comprehensiveness and accuracy of the data. Through cross-validation, contradictions and inconsistencies between data sources can be identified and corrected, thereby improving the consistency of data. This multi-dimensional verification can help build a more accurate credit assessment model. By integrating multi-dimensional data such as historical supply chain data, financial statements and credit reports, the gaps in data missing can be effectively filled to avoid credit assessment errors caused by incomplete information. By obtaining supply chain data features of consistent supply chain data, financial data features of consistent financial data and credit data features of consistent credit data, and performing dimensionality reduction processing on the supply chain data features, financial data features and credit data features respectively, the target supply chain features, target financial features and target credit features are correspondingly obtained. By obtaining consistent supply chain data , financial data and credit data, and comprehensively process their characteristics, so as to fully understand the credit status of the enterprise from multiple dimensions. This can not only combine the financial status of the enterprise, but also consider its supply chain stability and credit history, so as to provide a more comprehensive credit assessment. Through dimensionality reduction processing, the most representative and effective information can be extracted from a large number of features, redundant data and noise can be removed, and the analysis process can be simplified. The target supply chain characteristics, target financial characteristics and target credit characteristics obtained after dimensionality reduction can better capture the core credit risk information of the enterprise, thereby improving the efficiency and accuracy of the assessment model. The debt repayment ability index is calculated according to the target financial characteristics and target credit characteristics, the financial health index is calculated according to the target supply chain characteristics and target financial characteristics, and the credit risk index is obtained according to the target supply chain characteristics and target credit characteristics. The debt repayment ability index, financial health index and credit risk index are input into the credit risk prediction model to obtain the credit assessment value. By introducing multi-dimensional data such as financial characteristics, credit characteristics and supply chain characteristics, it is possible to make up for the omissions or deviations that may exist in a single data source, thereby more comprehensively reflecting the comprehensive situation of the enterprise and avoiding the misjudgment that may be caused by a single data source.Combining multiple indicators such as debt-paying ability, financial health, and credit risk can more accurately capture the true credit status of an enterprise, because these indicators reflect the financial health, operational stability, market performance, and potential risks of an enterprise from different angles. Therefore, evaluating an enterprise by integrating multi-dimensional characteristics can improve the accuracy of predictions and reduce misjudgments due to single or one-sided data. In traditional models, financial data is usually historical and may not be able to reflect the latest credit status of an enterprise in real time. By introducing dynamic data such as supply chain characteristics, the current operating status and changes of an enterprise can be captured more timely. By improving the timeliness of data, the credit changes of an enterprise can be more sensitively reflected, especially when facing uncertainties such as market fluctuations and supply chain disruptions. Potential risks can be predicted earlier. By calculating the debt-paying ability index, financial health index, and credit risk index, the credit risk of an enterprise can be analyzed from multiple angles, which can not only identify the current It can not only identify current financial problems, but also warn of possible future credit risks in advance, thereby providing more accurate credit assessments. Due to the introduction of a variety of different data sources, such as financial statements, supply chain information, credit reports, etc., even if a data source is missing, other data sources can still provide effective information supplements, thereby reducing the impact of single data missing on the overall analysis results, enhancing the model's fault tolerance for data missing and incompleteness, and improving the stability and reliability of the overall analysis. By inputting the debt repayment capacity index, financial health index and credit risk index into the credit risk prediction model, using these multi-dimensional and comprehensive information, the model can obtain more input data, thereby improving the model's prediction ability and robustness. This is because richer data features help to improve the accuracy and robustness of the credit risk prediction model. This method not only optimizes the breadth and depth of data use, but also improves the performance of the credit risk prediction model to a certain extent. ,

[0084] In one embodiment, the step S2 of cross-validating the historical supply chain data source, historical enterprise financial statements, and historical enterprise credit reports to obtain consistent supply chain data, consistent financial data, and consistent credit data includes:

[0085] S21, respectively obtaining all supply chain data from the historical supply chain data source, all financial data from historical enterprise financial statements, and all credit data from historical enterprise credit reports;

[0086] S22, extracting data identical to all financial data and all credit data from all supply chain data to obtain consistent supply chain data;

[0087] S23, extracting data identical to all supply chain data and all credit data from all the financial data to obtain consistent financial data;

[0088] S24. Extract the data identical to all the financial data and all the supply chain data from all the credit data to obtain consistent credit data.

[0089] As described in the above steps S21-S24, the present invention obtains all supply chain data from historical supply chain data sources, all financial data from historical corporate financial statements, and all credit data from historical corporate credit reports, and extracts the same data as all financial data and all credit data from all supply chain data to obtain consistent supply chain data, extracts the same data as all supply chain data and all credit data from all financial data to obtain consistent financial data, and extracts the same data as all financial data and all supply chain data from all credit data to obtain consistent credit data. By comparing data from three different fields of supply chain, finance, and credit and extracting consistent data, various types of information can be effectively integrated to comprehensively evaluate the credit risk of the enterprise from multiple dimensions. If only a single data source (such as financial data) is relied upon, risk information in the supply chain or potential problems in the credit report may be ignored, resulting in distorted credit evaluation results. By comparing across data sources, it is possible to Reduce the missing or bias in a single data source to ensure that the evaluation results are more accurate and comprehensive. By cross-comparing supply chain data, financial data and credit data, the consistency and reliability of the data can be verified. When different data sources are inconsistent in certain key indicators, it may indicate that the data is abnormal or risky. This consistency analysis helps to discover potential problems in the data and thus improve the quality and credibility of the data. This method can not only be analyzed based on historical data, but also provide support for real-time tracking of corporate credit. The credit status of an enterprise is dynamically changing, especially in today's increasingly globalized and information-based world. The credit status of an enterprise may be affected by factors such as supply chain changes, market fluctuations or financial strategy adjustments. By continuously comparing and updating data from different sources, changes in corporate credit can be reflected more quickly and flexibly. Through cross-validation of multi-dimensional data, the impact of this single point of failure can be reduced, making the evaluation results more robust and reducing the risks caused by errors in individual data sources.

[0090] In one embodiment, step S3 of performing dimensionality reduction processing on the supply chain data features to obtain target supply chain features includes:

[0091] S31, performing data standardization processing on the supply chain data characteristics to obtain standard supply chain characteristics;

[0092] S32, performing mean filling processing on the standard supply chain characteristics to obtain complete supply chain characteristics;

[0093] S33, obtaining a supply chain covariance matrix according to the complete supply chain characteristics, and obtaining all supply chain eigenvalues ​​and supply chain eigenvectors of the supply chain covariance matrix;

[0094] S34, marking the supply chain feature vectors corresponding to multiple supply chain feature values ​​greater than the supply chain threshold as principal component feature vectors;

[0095] S35, constructing a projection matrix according to the plurality of principal component eigenvectors;

[0096] S36. Perform a linear transformation on the supply chain covariance matrix according to the projection matrix, and project it onto the principal component space to obtain target supply chain characteristics.

[0097] As described in the above steps S31-S36, the steps of performing dimensionality reduction processing on financial data features to obtain target financial features and performing dimensionality reduction processing on credit data features to obtain target credit features are the same as the steps of performing dimensionality reduction processing on supply chain data features to obtain target supply chain features in this embodiment, and therefore are not described here. The present invention obtains standard supply chain features by performing data standardization processing on supply chain data features, and performs mean filling processing on standard supply chain features to obtain complete supply chain features. Traditional corporate credit assessment usually relies on a single data source such as financial statements and credit reports. These data sources may contain different metrics, data formats and dimensions, and have heterogeneity problems, resulting in the inability to conduct unified analysis. Data standardization converts data with different features into a unified scale, so that various data sources can be compared and comprehensively analyzed on the same platform, which can reduce errors caused by differences in data sources and enhance the model's adaptability to multi-source data. By performing mean filling processing on the data, missing values ​​can be effectively filled, thereby This method ensures the integrity of the data. Even if the data is incomplete, the stability and accuracy of the evaluation model can be maintained, avoiding the bias caused by missing data, enhancing the availability of data, and avoiding the negative impact of a large amount of data loss or bias on the analysis results. The supply chain covariance matrix is ​​obtained through complete supply chain characteristics, and all supply chain eigenvalues ​​and supply chain eigenvectors of the supply chain covariance matrix are obtained. By marking the supply chain eigenvectors corresponding to multiple supply chain eigenvalues ​​greater than the supply chain threshold as principal component eigenvectors, by adopting supply chain characteristics, it is possible to comprehensively capture all aspects of the enterprise's operations from multiple dimensions such as supply chain management, logistics, procurement, inventory, etc., thereby avoiding the problem of incomplete or distorted information that may be caused by a single data source. This improves the diversity and comprehensiveness of the data, thereby more accurately evaluating the credit status of the enterprise. In actual applications, many companies' financial statements or credit reports may have missing data, or the data quality is not high. These missing data or inaccurate information may cause deviations in credit assessment results.By using the eigenvalues ​​and eigenvectors of the covariance matrix for dimensionality reduction, it is possible to remove redundant information and capture the main factors of variation, thereby reducing the impact of noise and missing data on the evaluation results. In this way, effective credit information can still be extracted when the data is incomplete or of low quality, thereby improving the robustness of the evaluation. There is usually a certain correlation between the various characteristics in the supply chain. For example, the credit status of a supplier may be closely related to the financial health of the company. The characteristic analysis of the covariance matrix can reveal these potential correlations. By identifying the principal component eigenvectors of the covariance matrix, it can be revealed which characteristics have the greatest impact on the company's credit, thereby helping to reveal the inherent correlation between different variables, thereby improving the understanding of the company's overall credit status. By analyzing the covariance matrix of the supply chain and its characteristics, it is possible to comprehensively consider the impact of multiple factors on the basis of multi-dimensional data and improve the ability to predict corporate credit risk, especially in the face of a complex and changing market environment. It can provide a more accurate credit assessment. A projection matrix is ​​constructed through multiple principal component eigenvectors, and the supply chain covariance matrix is ​​linearly transformed according to the projection matrix and projected onto the principal component. The target supply chain characteristics are obtained in the sub-space, and principal component analysis is used to integrate multiple features such as financial data, market data, and supply chain data into a low-dimensional space. In this way, through the integration of multiple data sources, the evaluation model can more comprehensively capture the multidimensional characteristics of corporate credit and avoid the one-sidedness that may be caused by a single data source. Principal component analysis reduces the dimension of data by extracting the principal components in the data, removes redundant information, and focuses on the features that have the greatest impact on corporate credit risk assessment. This not only simplifies the calculation process, but also reduces the errors caused by data noise or redundancy, making the final evaluation results more accurate and efficient. Through principal component analysis, multiple data sources can be integrated and projected. Even if some data are missing, the supplementation of other data sources can ensure the comprehensiveness of corporate credit assessment. The possible impact of missing data can be inferred through the remaining relevant data, thereby improving the accuracy and robustness of the overall assessment. Through the linear transformation of multiple principal components, a more robust evaluation model can be obtained. Principal component analysis reduces the sensitivity to noise and outliers by concentrating the main change directions in the data, and enhances the stability of the model in the face of different companies or in long-term evaluations.

[0098] In one embodiment, the step S6 of calculating the solvency index according to the target financial characteristics and the target credit characteristics includes:

[0099] S61. Obtaining a debt repayment index and a profitability index according to the target financial characteristics, and obtaining a current ratio, a quick ratio and an interest coverage index according to the debt repayment index;

[0100] S62. Obtaining a debt coverage ratio according to the current ratio, quick ratio and interest coverage index;

[0101] S63, obtaining a net profit margin and a return on net assets according to the profitability indicator, and obtaining a profitability ratio according to the net profit margin and the return on net assets;

[0102] S64. Acquire a credit utilization rate and a debt compliance rate according to the target credit characteristics;

[0103] S65. Calculate the debt repayment capacity index according to the credit utilization rate, debt compliance rate, debt coverage ratio and profitability ratio, wherein the calculation formula is:

[0104]

[0105] Among them, C(ZN) represents the debt repayment ability index, X(LV) represents the credit utilization rate, Z(YL) represents the debt fulfillment rate, Z(FB) represents the debt coverage ratio, and Y(BL) represents the profitability ratio.

[0106] As described in the above steps S61-S65, the present invention obtains debt repayment indicators and profitability indicators according to the target financial characteristics, and obtains the current ratio, quick ratio and interest coverage index according to the debt repayment indicator, obtains the debt coverage ratio through the current ratio, quick ratio and interest coverage index, obtains the net profit margin and return on net assets through the profitability indicator, and obtains the profitability ratio according to the net profit margin and return on net assets, wherein the current ratio is the ratio of current assets to current liabilities, which is used to measure the short-term debt repayment ability of an enterprise, the quick ratio is the ratio of assets after deducting inventory from current assets to current liabilities, and reflects the guarantee of the more liquid part of the enterprise for short-term debt repayment, and the profitability ratio usually refers to the net profit to total income, assets or shareholders' equity. The debt coverage ratio measures the profitability of a company. The debt coverage ratio indicates the proportion of a company's debt to its total assets. The credit utilization ratio usually refers to the ratio of a company's debt to its available credit line. The debt service ratio measures the company's ability to pay interest and principal with its profitability. By comprehensively considering multiple indicators such as debt repayment indicators, profitability indicators, current ratios, quick ratios, and interest coverage indexes, a company's credit status can be more comprehensively assessed, reducing the deviation or distortion caused by the limitations of a single data source. This multi-dimensional assessment can more truly reflect the company's financial health. By correlating core indicators such as debt repayment ability and profitability, it can effectively reflect whether the company has sufficient financial resources to cope with future financial burdens. For example, companies with insufficient debt repayment and profitability may face higher credit risks in the future. The addition of current ratio and quick ratio can further confirm whether the company has sufficient short-term debt repayment ability, while the interest coverage ratio focuses on the company's debt burden and payment ability. This improvement in risk prediction ability helps to more accurately assess and predict the company's credit risk. By obtaining debt repayment indicators and profitability indicators based on target financial characteristics, and further deriving other key indicators through these indicators, it can help overcome the problems of missing data and reliance on a single data source that may occur in traditional corporate credit assessments, and can improve the comprehensiveness, accuracy, risk prediction ability and decision-making quality of the assessment, thereby better reflecting the company's credit status. Effectively support investment decisions and risk management. Obtain credit utilization and debt fulfillment rate through target credit characteristics. Target credit characteristics form a comprehensive evaluation by combining multiple indicators such as credit utilization, debt fulfillment rate, debt coverage ratio and profitability ratio, which can reduce the problem of incomplete or distorted single data source. In this way, the evaluation results can cover the company's financial health, debt repayment ability and profitability and other information. Through the combination of multi-dimensional credit characteristics such as credit utilization rate and debt fulfillment rate, the evaluation model can complement different data sources. If a certain type of data is missing or inaccurate, other data can still provide strong support for the evaluation, thereby reducing the potential risks caused by the lack of a single data. The above indicators are used in combination for calculation.It can improve the meticulousness of the assessment results, thereby more accurately identifying potential problems that may exist in the company's debt repayment ability, debt risk, etc., obtain credit utilization rate, debt fulfillment rate, debt coverage ratio and profitability ratio through target credit characteristics, and calculate the debt repayment ability index through these indicators. It can comprehensively assess the company's credit status from multiple dimensions, avoiding the problem of traditional credit assessment being too dependent on a single data source, and can more accurately and comprehensively identify the company's credit risk, improve the accuracy and reliability of the assessment, and thus improve risk management capabilities.

[0107] In one embodiment, the step S7 of calculating the financial health index according to the target supply chain characteristics and the target financial characteristics includes:

[0108] S71, acquiring inventory management information and supplier management information according to the target supply chain characteristics, and acquiring inventory turnover rate and inventory turnover cycle according to the inventory management information;

[0109] S72. Obtaining the ratio of accounts payable to accounts receivable and the ratio of purchases to sales according to the supplier management information;

[0110] S73, obtaining supply chain management efficiency according to the inventory turnover rate, inventory turnover cycle, accounts payable to accounts receivable ratio and purchase to sales ratio;

[0111] S74, obtaining operation efficiency information and cash flow information according to the target financial characteristics, and obtaining inventory turnover rate and accounts receivable turnover rate according to the operation efficiency information;

[0112] S75, obtaining a cash flow ratio and current liabilities according to the cash flow information, and obtaining an operating cash flow according to the cash flow ratio and current liabilities;

[0113] S76, obtaining a debt ratio according to the current liabilities, and obtaining working capital according to the operating cash flow and current liabilities;

[0114] S77. Calculate the financial liquidity value based on the debt ratio, operating cash flow, working capital, inventory turnover rate and accounts receivable turnover rate, wherein the calculation formula is:

[0115]

[0116] Among them, C(LD) represents financial liquidity value, J(XJ) represents operating cash flow, Y(YZ) represents working capital, C(ZL) represents inventory turnover rate, Y(SL) represents accounts receivable turnover rate, and F(BL) represents debt ratio;

[0117] S78. Obtain a financial health index based on the financial liquidity value and supply chain management efficiency.

[0118] As described in the above steps S71-S78, the present invention obtains inventory management information and supplier management information through target supply chain characteristics, obtains inventory turnover rate and inventory turnover cycle based on inventory management information, obtains accounts payable to accounts receivable ratio and purchase to sales ratio through supplier management information, and obtains supply chain management efficiency through inventory turnover rate, inventory turnover cycle, accounts payable to accounts receivable ratio and purchase to sales ratio, wherein the inventory turnover rate indicates the frequency with which the enterprise inventory is sold or used, that is, the number of inventory turnovers within a certain period of time, the inventory turnover cycle reflects the entire cycle time from purchasing raw materials to selling products, that is, the time required for inventory turnover, and the accounts receivable to accounts payable ratio reflects the enterprise The degree of capital utilization in the sales and procurement links. The procurement-to-sales ratio reflects the relationship between the company's procurement costs and sales revenue, and is usually used to measure the company's operating efficiency and profitability. Traditional credit assessment methods usually rely on financial data in financial statements. These data reflect the company's financial status, but often ignore important factors such as supply chain operations, inventory management and supplier management. By combining supply chain characteristics such as inventory turnover rate, inventory turnover cycle, accounts payable to accounts receivable ratio, procurement-to-sales ratio into the credit assessment model, the insufficiency of financial data can be effectively supplemented to obtain a more comprehensive and accurate risk assessment result. Information such as inventory management and supplier management is dynamic and can reflect the daily operations of the company. The actual situation of operations. This information can usually be obtained outside of financial reports. Therefore, they can effectively fill the lags and incompleteness that may appear in financial statements, and further improve the timeliness and accuracy of evaluation. The level of inventory management directly affects the capital occupation. The inventory turnover rate and inventory turnover cycle can reflect the efficiency of the company's capital use and its ability to respond to market demand. If a company's inventory turnover rate is low or the inventory turnover cycle is long, it may mean that the product is unsalable or overproduced, thus causing cash flow problems. A high proportion of accounts payable or accounts receivable may mean that the company is facing greater financial pressure, or that the credit status of suppliers and customers is poor. These indicators can help identify liquidity problems. Providing richer background information for credit risk assessment, combined with features such as inventory management and supplier management, helps to formulate more accurate credit assessment strategies based on the specific circumstances of different companies, thereby optimizing the company's credit management decisions, obtaining operational efficiency information and cash flow information through target financial characteristics, and obtaining inventory turnover and accounts receivable turnover based on operational efficiency information, obtaining cash flow ratios and current liabilities through cash flow information, and obtaining operating cash flow based on cash flow ratios and current liabilities, obtaining debt ratios through current liabilities, and obtaining working capital based on operating cash flow and current liabilities. Traditional credit assessment methods usually rely on a single data source, such as financial statements, credit reports, or bank credit scores. The above methods take into account multiple aspects of corporate finance, such as cash flow, inventory management, accounts receivable, etc.It comprehensively reflects the actual operation of the enterprise and reduces the deviation that may be caused by a single data source. By combining multi-dimensional information such as operating efficiency, cash flow and liabilities, it can fill the data gaps that may exist in traditional methods, improve the integrity of the data, and make the evaluation results more comprehensive and accurate. Indicators such as cash flow information and operating efficiency can reflect the short-term debt repayment ability and operational health of the enterprise, which may not be immediately reflected in traditional financial statements. Combining indicators such as cash flow, accounts receivable and inventory turnover can better capture potential risks in the operation of the enterprise, and further evaluate the debt level and debt repayment ability of the enterprise through cash flow ratio and current liabilities. This method can help the evaluator deeply understand the operating efficiency of the enterprise under different financial conditions. The combination of debt ratio and operating cash flow can deeply analyze the debt structure and debt repayment ability of the enterprise, especially when the working capital management and cash flow are insufficient. It avoids the misjudgment caused by relying on debt ratio or cash flow alone. The working capital situation can be obtained by calculating operating cash flow and current liabilities, which can more accurately reflect the capital operation and operation ability of the enterprise. This can more truly reflect the financial health of the enterprise than relying solely on the data in the income statement. The financial liquidity value is calculated by debt ratio, operating cash flow, working capital, inventory turnover rate and accounts receivable turnover rate, and then the financial health index is obtained through financial liquidity value and supply chain management efficiency. Among them, the debt ratio measures the proportion of total liabilities to total assets of the enterprise, reflecting the enterprise's Financial leverage and debt repayment pressure. A higher debt ratio may mean that the company has greater financial risks and faces a higher risk of default. Operating cash flow is a measure of the company's ability to actually generate cash flow, reflecting the cash inflows and outflows in the company's operating activities. Healthy operating cash flow can ensure that the company can pay short-term debts and supply chain payments, reflecting the health of the company's liquidity. Working capital is a measure of the company's available working capital in daily operations. The more working capital, the more capable the company is to cope with the capital needs in daily operations and reduce credit risk. Inventory turnover rate is a measure of the turnover efficiency of the company's inventory. A high inventory turnover rate means that the company can sell inventory quickly, reduce the risk of inventory backlogs, and increase working capital. Accounts receivable turnover rate is a measure of the company's accounts receivable recovery. Speed, a higher accounts receivable turnover rate means that the company can quickly collect payments, increase cash flow and reduce bad debt risks. By combining multiple financial indicators such as debt ratio, operating cash flow, working capital, and operational data such as supply chain management efficiency, a more comprehensive and three-dimensional financial health portrait can be obtained, avoiding the deviation caused by relying on a single data source. By using multiple financial ratios and operational efficiency indicators, the company's true financial status and operational health can be more accurately reflected. These indicators can not only reveal the profitability of the company, but also the company's liquidity, debt repayment ability and capital turnover efficiency. By incorporating operational indicators such as inventory turnover rate and accounts receivable turnover rate into the evaluation model, potential problems that may arise during the company's operations can be reflected in a timely manner.By combining financial liquidity value with supply chain efficiency, the operational risks that an enterprise may face can be discovered earlier. Therefore, calculating the financial health index through the above-mentioned comprehensive method can effectively solve the problems of incomplete and incomplete data in traditional credit assessment methods, and provide more accurate, real-time and comprehensive credit risk assessment. With the support of data from multiple dimensions such as financial liquidity value and operational efficiency, credit assessment can reflect the comprehensive financial status and operational risks of an enterprise.

[0119] In one embodiment, the step S8 of acquiring the credit risk index according to the target supply chain characteristics and the target credit characteristics includes:

[0120] S81, obtaining supply chain performance data according to the target supply chain characteristics, and obtaining supply chain dependence and supply chain stability according to the supply chain performance data;

[0121] S82, obtaining a delivery on-time rate and a delivery quality rate according to the supply chain stability, and obtaining a timeliness coefficient according to the delivery on-time rate and the supply chain dependence;

[0122] S83. Obtaining a quality compliance coefficient according to the supply chain dependence and the delivery quality rate;

[0123] S84. Obtaining a credit score and a default risk score according to the target credit characteristics;

[0124] S85. Obtain a credit risk index according to the timeliness coefficient, compliance coefficient, credit score and default risk score.

[0125] As described in the above steps S81-S85, the present invention obtains supply chain performance data through target supply chain characteristics, and obtains supply chain dependence and supply chain stability based on the supply chain performance data, obtains delivery punctuality and delivery quality rate through supply chain stability, and obtains timeliness coefficient based on delivery punctuality and supply chain dependence, and obtains quality compliance coefficient through supply chain dependence and delivery quality rate, wherein the timeliness coefficient measures the ability of the supply chain to complete delivery within the agreed time, directly reflecting the delivery punctuality of the supply chain, and the quality compliance coefficient measures whether the product or service delivered by the supply chain meets the predetermined quality standards, and comprehensively considers the performance data of various aspects of the supply chain. Supply chain performance data can comprehensively evaluate the performance of enterprises at multiple levels, thereby obtaining a more comprehensive and accurate credit assessment. Financial statements and credit reports are usually lagging and data update is slow, while supply chain performance data can often provide more timely and dynamic enterprise operation status. By real-time monitoring and evaluating the stability, delivery quality and timeliness of the supply chain, it can provide early warning before the enterprise encounters risks, thereby providing more timely credit assessment and avoiding relying solely on outdated financial data to make decisions. Through indicators such as supply chain stability, dependence, and delivery punctuality, the operational risks of the enterprise can be effectively assessed. The fragility or instability of the supply chain is often an important factor affecting the operational efficiency and The credit score and default risk score are obtained through the target credit characteristics, and the credit risk index is obtained through the timeliness coefficient, compliance coefficient, credit score and default risk score. By integrating multi-dimensional data such as target credit characteristics, timeliness coefficient, compliance coefficient, etc., not only the financial status of the enterprise is considered, but also other key factors (such as market environment, management quality, industry trends, etc.) can be included to provide a more comprehensive and accurate credit assessment. This multi-dimensional assessment method can effectively supplement the shortcomings of traditional methods and reduce the limitations of a single data source. By adding timeliness coefficient and compliance coefficient, it can ensure that the assessment model can better reflect the current situation of the enterprise. The actual credit status and timeliness coefficient can ensure the updating frequency of the evaluation data, so that the model can adapt to the rapid changes in the market. The compliance coefficient can help evaluate the impact of the company's compliance behavior on credit risk, thereby capturing more possible risk points. Through the comprehensive credit score, default risk score, timeliness coefficient and compliance coefficient, a more comprehensive and multi-dimensional risk assessment system can be constructed to reduce deviations and improve the robustness and stability of the scoring system. Through the introduction of the timeliness coefficient, credit assessment can more quickly reflect changes in corporate credit, especially when the external environment changes drastically (such as economic crises, industry fluctuations, etc.), and can promptly reflect the process of risk accumulation.In this way, the company's default risk or credit deterioration trend can be predicted in advance, early warnings can be issued in time, and decision makers can be helped to take appropriate risk response measures. The compliance of an enterprise can not only reflect the stability of its operations, but also reflect potential legal and policy risks. Through the application of compliance coefficients, the accuracy of corporate credit assessments can be further improved, helping to identify potential compliance risks, thereby identifying potential risk points that may lead to corporate defaults in advance. Traditional credit assessments usually rely on financial indicators and credit history, which may lead to insufficient predictions of future risks. The comprehensive use of multiple credit scores, default risk scores, timeliness, compliance and other factors can help decision makers understand the company's potential risks more clearly and enhance the scientific nature and foresight of decision-making. This method not only improves the comprehensiveness of corporate credit assessments, but also better supports risk management decisions and helps financial institutions and other companies make more informed decisions in a changing market environment.

[0126] The present application also provides a credit data collection system based on deep learning, comprising:

[0127] A first acquisition module is used to acquire historical data features of enterprise credit data, wherein the historical data features include historical supply chain data sources, historical enterprise financial statements and historical enterprise credit reports;

[0128] A verification module, used to cross-verify the historical supply chain data source, historical enterprise financial statements and historical enterprise credit reports to obtain consistent supply chain data, consistent financial data and consistent credit data;

[0129] A first dimensionality reduction module is used to obtain supply chain data features of the consistent supply chain data, and perform dimensionality reduction processing on the supply chain data features to obtain target supply chain features;

[0130] A second dimension reduction module is used to obtain the financial data features of the consistent financial data and perform dimension reduction processing on the financial data features to obtain target financial features;

[0131] A third dimension reduction module is used to obtain the credit data features of the consistent credit data, and perform dimension reduction processing on the credit data features to obtain target credit features;

[0132] A first calculation module, used to calculate a debt repayment ability index according to the target financial characteristics and target credit characteristics;

[0133] A second calculation module, used to calculate a financial health index according to the target supply chain characteristics and the target financial characteristics;

[0134] The second acquisition module is used to obtain a credit risk index based on the target supply chain characteristics and target credit characteristics, and input the debt repayment ability index, financial health index and credit risk index into a credit risk prediction model to obtain a credit assessment value.

[0135] In one embodiment, the second acquisition module includes:

[0136] A first acquisition unit is used to acquire supply chain performance data according to the target supply chain characteristics, and acquire supply chain dependence and supply chain stability according to the supply chain performance data;

[0137] A second acquisition unit is used to acquire a delivery on-time rate and a delivery quality rate according to the supply chain stability, and to acquire a timeliness coefficient according to the delivery on-time rate and the supply chain dependence;

[0138] A third acquisition unit is used to acquire a quality compliance coefficient according to the supply chain dependence and the delivery quality rate;

[0139] A fourth acquisition unit, configured to acquire a credit score and a default risk score according to the target credit feature;

[0140] The fifth acquisition unit is used to acquire a credit risk index according to the timeliness coefficient, the compliance coefficient, the credit score and the default risk score.

[0141] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned credit data collection method based on deep learning when executing the computer program.

[0142] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned credit data collection method based on deep learning.

[0143] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0144] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0145] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A credit data collection method based on deep learning, characterized in that: include: Acquiring historical data features of enterprise credit data, wherein the historical data features include historical supply chain data sources, historical enterprise financial statements, and historical enterprise credit reports; Cross-validating the historical supply chain data source, historical enterprise financial statements, and historical enterprise credit reports to obtain consistent supply chain data, consistent financial data, and consistent credit data; Acquire supply chain data features of the consistent supply chain data, and perform dimensionality reduction processing on the supply chain data features to obtain target supply chain features; Acquiring financial data features of the consistent financial data, and performing dimensionality reduction processing on the financial data features to obtain target financial features; Acquiring credit data features of the consistent credit data, and performing dimensionality reduction processing on the credit data features to obtain target credit features; Calculate a solvency index based on the target financial characteristics and target credit characteristics; Calculating a financial health index based on the target supply chain characteristics and the target financial characteristics; A credit risk index is obtained according to the target supply chain characteristics and the target credit characteristics, and the debt repayment capacity index, the financial health index and the credit risk index are input into a credit risk prediction model to obtain a credit assessment value.

2. The credit data collection method based on deep learning according to claim 1, characterized in that: The step of cross-validating the historical supply chain data source, historical enterprise financial statements, and historical enterprise credit reports to obtain consistent supply chain data, consistent financial data, and consistent credit data includes: Respectively obtain all supply chain data of the historical supply chain data source, all financial data of historical enterprise financial statements, and all credit data of historical enterprise credit reports; Extracting data identical to all financial data and all credit data from all supply chain data to obtain consistent supply chain data; Extracting data identical to all supply chain data and all credit data from all financial data to obtain consistent financial data; The data identical to those in all the financial data and all the supply chain data are extracted from all the credit data to obtain consistent credit data.

3. The credit data collection method based on deep learning according to claim 1, characterized in that: The step of performing dimensionality reduction processing on the supply chain data features to obtain target supply chain features includes: Performing data standardization processing on the supply chain data characteristics to obtain standard supply chain characteristics; Performing mean filling processing on the standard supply chain characteristics to obtain complete supply chain characteristics; Obtaining a supply chain covariance matrix according to the complete supply chain characteristics, and obtaining all supply chain eigenvalues ​​and supply chain eigenvectors of the supply chain covariance matrix; Mark the supply chain feature vectors corresponding to multiple supply chain feature values ​​greater than the supply chain threshold as the principal component feature vectors; Constructing a projection matrix according to a plurality of the principal component eigenvectors; The supply chain covariance matrix is ​​linearly transformed according to the projection matrix and projected onto the principal component space to obtain the target supply chain characteristics.

4. The credit data collection method based on deep learning according to claim 1, characterized in that: The step of calculating the debt repayment ability index according to the target financial characteristics and the target credit characteristics comprises: Obtaining debt repayment indicators and profitability indicators based on the target financial characteristics, and obtaining current ratio, quick ratio and interest coverage index based on the debt repayment indicators; Obtain the debt coverage ratio based on the current ratio, quick ratio and interest coverage index; Obtaining a net profit margin and a return on net assets based on the profitability indicator, and obtaining a profitability ratio based on the net profit margin and the return on net assets; Obtaining a credit utilization rate and a debt compliance rate based on the target credit characteristics; The debt repayment ability index is calculated based on the credit utilization ratio, debt compliance ratio, debt coverage ratio and profitability ratio, wherein the calculation formula is: Among them, C(ZN) represents the debt repayment ability index, X(LV) represents the credit utilization rate, Z(YL) represents the debt fulfillment rate, Z(FB) represents the debt coverage ratio, and Y(BL) represents the profitability ratio.

5. The credit data collection method based on deep learning according to claim 1, characterized in that: The step of calculating the financial health index according to the target supply chain characteristics and the target financial characteristics comprises: Acquire inventory management information and supplier management information according to the target supply chain characteristics, and acquire inventory turnover rate and inventory turnover cycle according to the inventory management information; Obtaining accounts payable to accounts receivable ratios and purchases to sales ratios based on the supplier management information; Obtaining supply chain management efficiency based on the inventory turnover ratio, inventory turnover cycle, accounts payable to accounts receivable ratio and purchase to sales ratio; Acquire operation efficiency information and cash flow information according to the target financial characteristics, and acquire inventory turnover rate and accounts receivable turnover rate according to the operation efficiency information; Obtaining a cash flow ratio and current liabilities based on the cash flow information, and obtaining operating cash flow based on the cash flow ratio and current liabilities; Obtain the debt ratio based on the current liabilities, and obtain the working capital based on the operating cash flow and current liabilities; The financial liquidity value is calculated based on the debt ratio, operating cash flow, working capital, inventory turnover rate and accounts receivable turnover rate, where the calculation formula is: Among them, C(LD) represents financial liquidity value, J(XJ) represents operating cash flow, Y(YZ) represents working capital, C(ZL) represents inventory turnover rate, Y(SL) represents accounts receivable turnover rate, and F(BL) represents debt ratio; A financial health index is obtained according to the financial liquidity value and the supply chain management efficiency.

6. The credit data collection method based on deep learning according to claim 1, characterized in that: The step of obtaining a credit risk index according to the target supply chain characteristics and the target credit characteristics comprises: Acquire supply chain performance data according to the target supply chain characteristics, and acquire supply chain dependence and supply chain stability according to the supply chain performance data; Obtaining a delivery on-time rate and a delivery quality rate according to the supply chain stability, and obtaining a timeliness coefficient according to the delivery on-time rate and the supply chain dependence; Obtaining a quality compliance coefficient based on the supply chain dependency and the delivery quality rate; Obtaining a credit score and a default risk score based on the target credit characteristics; A credit risk index is obtained based on the timeliness coefficient, compliance coefficient, credit score and default risk score.

7. A credit data collection system based on deep learning, characterized in that: include: A first acquisition module is used to acquire historical data features of enterprise credit data, wherein the historical data features include historical supply chain data sources, historical enterprise financial statements and historical enterprise credit reports; A verification module, used to cross-verify the historical supply chain data source, historical enterprise financial statements and historical enterprise credit reports to obtain consistent supply chain data, consistent financial data and consistent credit data; A first dimensionality reduction module is used to obtain supply chain data features of the consistent supply chain data, and perform dimensionality reduction processing on the supply chain data features to obtain target supply chain features; A second dimension reduction module is used to obtain the financial data features of the consistent financial data and perform dimension reduction processing on the financial data features to obtain target financial features; A third dimension reduction module is used to obtain the credit data features of the consistent credit data, and perform dimension reduction processing on the credit data features to obtain target credit features; A first calculation module, used to calculate a debt repayment ability index according to the target financial characteristics and target credit characteristics; A second calculation module, used to calculate a financial health index according to the target supply chain characteristics and the target financial characteristics; The second acquisition module is used to obtain a credit risk index based on the target supply chain characteristics and target credit characteristics, and input the debt repayment ability index, financial health index and credit risk index into a credit risk prediction model to obtain a credit assessment value.

8. The credit data collection system based on deep learning according to claim 7, characterized in that: The second acquisition module includes: A first acquisition unit is used to acquire supply chain performance data according to the target supply chain characteristics, and acquire supply chain dependence and supply chain stability according to the supply chain performance data; A second acquisition unit is used to acquire a delivery on-time rate and a delivery quality rate according to the supply chain stability, and to acquire a timeliness coefficient according to the delivery on-time rate and the supply chain dependence; A third acquisition unit is used to acquire a quality compliance coefficient according to the supply chain dependence and the delivery quality rate; A fourth acquisition unit, configured to acquire a credit score and a default risk score according to the target credit feature; The fifth acquisition unit is used to acquire a credit risk index according to the timeliness coefficient, the compliance coefficient, the credit score and the default risk score.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

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

Images