Credit decision-making method and device and storage medium
Through entropy weight method and data mining technology, the credit risk model is optimized, and the problem of unreasonable lending strategies for small and medium-sized enterprises is solved, and more scientific and accurate loan decisions are achieved.
Patent Information
- Application Number
- CN202310637534.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-31
- Publication Date
- 2024-12-03
AI Technical Summary
When banks face loan requests for small and medium-sized enterprises, there are fewer reference indicators in the existing technology and unscientific weight allocation, resulting in unreasonable lending strategies.
The entropy weight method is used to calculate the weight of the enterprise's quantitative index value, and linear regression is performed in combination with data mining software to optimize the credit risk model to ensure the objectivity of the loan strategy and the accuracy of prediction.
By quantitative calculation, the rationality and prediction accuracy of the loan strategy are improved, making the bank's loan strategies for different companies more in line with their debt repayment ability.
Smart Images

Figure CN119067764A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and in particular to a credit decision-making method, device and storage medium. Background Art
[0002] With the development of social economy, more and more small and medium-sized enterprises are being established, and at the same time, their loan requests from banks are also increasing.
[0003] In the prior art, when banks face loan requests from small and medium-sized enterprises, they make credit decisions based on national policies, enterprise transaction invoices, and the enterprise's upstream and downstream supply and demand parties. This credit decision is based on the subjective evaluation of the indicator problem by humans, and the weight of each reference indicator is determined. Then, based on the weight of each reference indicator, the final loan amount for the enterprise is determined.
[0004] However, the above scheme adopts fewer reference indicators, and the weight of each reference indicator is determined based on human subjective judgment, which makes the weight distribution of each reference indicator unscientific, resulting in unreasonable lending strategies of banks to enterprises. Summary of the invention
[0005] The present application provides a credit decision method, device and storage medium, which can solve the problem of unreasonable lending strategies of banks to different enterprises.
[0006] In order to achieve the above objectives, this application adopts the following technical solutions:
[0007] In a first aspect, the present application provides a credit decision-making method, which includes: using the entropy weight method to perform weight calculation on the quantitative indicator values of an enterprise to obtain the weights of the above-mentioned quantitative indicator values, and the above-mentioned quantitative indicator values are used to characterize the credit risk situation of the enterprise; according to the above-mentioned quantitative indicator values and the weights of the above-mentioned quantitative indicator values, the credit risk value of the enterprise is calculated; using data mining software, linear regression is performed on the credit rating and annual loan interest rate of the enterprise to obtain a first linear programming model, and the first linear programming model is used to characterize the relationship between the above-mentioned credit rating and the above-mentioned annual loan interest rate; according to the above-mentioned credit risk value, the above-mentioned first linear programming model is optimized to obtain a target linear programming model, and the prediction accuracy of the target linear programming model is greater than the prediction accuracy of the above-mentioned first linear programming model; the credit rating of the target enterprise is input into the above-mentioned target linear programming model for prediction to obtain the loan strategy of the above-mentioned target enterprise.
[0008] Based on the above technical solution, the credit decision method provided by the embodiment of the present application can use the entropy weight method to calculate the weight of the quantitative index value of the enterprise; then according to the quantitative index value and the weight of the quantitative index value, the credit risk value of the enterprise is calculated, and then the data mining software is used to perform linear regression on the credit rating and annual loan interest rate of the enterprise to obtain a first linear programming model, and the credit risk value calculated above is used to optimize the first linear programming model to obtain a target linear programming model, and finally the credit rating of the enterprise that needs a loan is input into the target linear programming model to obtain the bank's loan strategy for the enterprise. Since the calculation method of the above entropy weight method is a quantitative calculation method, it effectively avoids the influence of the user's subjective qualitative judgment on the weight of each quantitative index value, and ensures the objectivity and accuracy of the weight of each quantitative index value; and in the subsequent process of establishing the model, the credit risk value that can reflect the debt repayment ability of each enterprise is introduced to optimize the model, which improves the prediction accuracy of the model, so that the final determined loan strategy is more in line with the actual debt repayment ability of the enterprise, thereby ensuring the rationality of the bank's loan strategy for different enterprises.
[0009] In a first possible implementation manner of the first aspect, the above-mentioned enterprises include enterprises with credit records and enterprises without credit records, and the above-mentioned quantitative index values include a first quantitative index value and a second quantitative index value; before the above-mentioned entropy weight method is used to perform weight calculation on the quantitative index values of the enterprises, the above-mentioned method also includes: when the above-mentioned enterprises are enterprises with credit records, obtaining the first quantitative index value of the above-mentioned enterprises with credit records, and the first quantitative index value includes: the average price and tax of input invoices, the standard deviation of input invoices, the average price and tax of output invoices, the standard deviation of output invoices, the proportion of invalid and negative invoices, the frequency of upstream enterprises, the frequency of downstream enterprises and the credit rating; when the above-mentioned enterprises are enterprises without credit records, obtaining the first quantitative index value and the second quantitative index value of the above-mentioned enterprises without credit records, and the second quantitative index value includes: the average value of input amount, the average value of output amount, the average value of input tax, the average value of output tax, the average value of input price and tax and the average value of output price and tax.
[0010] In a second possible implementation of the first aspect, before obtaining the first quantitative index value of the enterprise with no credit record, the method further includes: in the case that the enterprise is an enterprise with no credit record, obtaining a second quantitative index value and a credit rating of a target enterprise with a credit record; wherein the second quantitative index value of the target enterprise with a credit record matches the second quantitative index value of the enterprise with no credit record.
[0011] In a third possible implementation of the first aspect, the obtaining of the first quantitative index value of the above-mentioned enterprise with no credit record includes: using a preset neural network to learn the correlation between the second quantitative index value of the above-mentioned target enterprise with a credit record and the credit rating, and obtaining a target neural network model; inputting the second quantitative index value of the above-mentioned enterprise with no credit record into the above-mentioned target neural network model for prediction, and obtaining the credit rating of the above-mentioned enterprise with no credit record.
[0012] In a fourth possible implementation of the first aspect, the credit risk value of the enterprise is calculated based on the quantitative index value and the weight of the quantitative index value, including: screening out the proportion of the invalidated and negative invoices in the first quantitative index value of the enterprise to obtain a third quantitative index value; and taking a weighted sum of the third quantitative index value and the proportion of the invalidated and negative invoices according to the weight of the third quantitative index value and the proportion of the invalidated and negative invoices to obtain the credit risk value of the enterprise.
[0013] In a second aspect, the present application provides a credit decision-making device, which includes: a calculation unit, a processing unit and a prediction unit.
[0014] Among them, the above-mentioned calculation unit is used to use the entropy weight method to perform weight calculation on the quantitative indicator value of the enterprise to obtain the weight of the above-mentioned quantitative indicator value, and the above-mentioned quantitative indicator value is used to characterize the credit risk situation of the enterprise; the above-mentioned calculation unit is also used to calculate the credit risk value of the enterprise based on the above-mentioned quantitative indicator value and the weight of the above-mentioned quantitative indicator value; the above-mentioned processing unit is used to use data mining software to perform linear regression on the credit rating and annual loan interest rate of the enterprise to obtain a first linear programming model, and the first linear programming model is used to characterize the relationship between the above-mentioned credit rating and the above-mentioned annual loan interest rate; the above-mentioned processing unit is also used to optimize the above-mentioned first linear programming model according to the above-mentioned credit risk value to obtain a target linear programming model, and the prediction accuracy of the target linear programming model is greater than the prediction accuracy of the above-mentioned first linear programming model; the above-mentioned prediction unit is used to input the credit rating of the target enterprise into the above-mentioned target linear programming model obtained by the processing unit for prediction, and obtain the loan strategy of the above-mentioned target enterprise.
[0015] In a first possible implementation manner of the second aspect, the above-mentioned enterprises include enterprises with credit records and enterprises without credit records, the above-mentioned quantitative index values include a first quantitative index value and a second quantitative index value, and the above-mentioned device also includes: an acquisition unit, wherein: the above-mentioned acquisition unit is used to: before the calculation unit adopts the entropy weight method to perform weight calculation on the quantitative index value of the enterprise, if the above-mentioned enterprise is an enterprise with a credit record, obtain the first quantitative index value of the enterprise with a credit record, and the first quantitative index value includes: the average price and tax of input invoices, the standard deviation of input invoices, the average price and tax of output invoices, the standard deviation of output invoices, the proportion of invalid and negative invoices, the frequency of upstream enterprises, the frequency of downstream enterprises and the credit rating; if the above-mentioned enterprise is an enterprise without a credit record, obtain the first quantitative index value and the second quantitative index value of the enterprise without a credit record, and the second quantitative index value includes: the average value of input amount, the average value of output amount, the average value of input tax, the average value of output tax, the average value of input price and tax and the average value of output price and tax.
[0016] In a second possible implementation of the second aspect, the acquisition unit is further used to acquire, before acquiring the first quantitative index value of the enterprise with no credit record, a second quantitative index value and a credit rating of a target enterprise with a credit record, in the case that the enterprise is an enterprise with no credit record; wherein the second quantitative index value of the target enterprise with a credit record matches the second quantitative index value of the enterprise with no credit record.
[0017] In a third possible implementation of the second aspect, the acquisition unit is specifically used to: use a preset neural network to learn the correlation between the second quantitative indicator value of the target enterprise with a credit record and the credit rating, and obtain a target neural network model; input the second quantitative indicator value of the enterprise without a credit record into the target neural network model for prediction, and obtain the credit rating of the enterprise without a credit record.
[0018] In a fourth possible implementation of the second aspect, the calculation unit is specifically used to: screen out the proportion of the above-mentioned invalidated and negative invoices in the first quantitative index value of the above-mentioned enterprise to obtain a third quantitative index value; and perform weighted summation of the third quantitative index value and the proportion of the above-mentioned invalidated and negative invoices according to the weights of the third quantitative index value and the proportion of the above-mentioned invalidated and negative invoices to obtain the credit risk value of the above-mentioned enterprise.
[0019] In a third aspect, the present application provides a credit decision device, comprising: a processor and a communication interface; the communication interface and the processor are coupled, and the processor is used to run a computer program or instructions to implement the credit decision method described in the first aspect and any possible implementation method of the first aspect.
[0020] In a fourth aspect, the present application provides a computer-readable storage medium, in which instructions are stored. When the instructions are executed on a terminal, the terminal executes the credit decision method described in the first aspect and any possible implementation of the first aspect.
[0021] In a fifth aspect, an embodiment of the present application provides a computer program product comprising instructions. When the computer program product is run on a credit decision device, the credit decision device executes the credit decision method as described in the first aspect and any possible implementation of the first aspect.
[0022] In a sixth aspect, an embodiment of the present application provides a chip, the chip including a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run a computer program or instructions to implement the credit decision method described in the first aspect and any possible implementation method of the first aspect.
[0023] Specifically, the chip provided in the embodiment of the present application also includes a memory for storing computer programs or instructions. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] Figure 1 A method flow chart of a credit decision method provided in an embodiment of the present application;
[0025] Figure 2 A credit risk analysis flow chart provided in an embodiment of the present application;
[0026] Figure 3 One of the indicator weight bar charts of a credit decision method provided in an embodiment of the present application;
[0027] Figure 4 A training flow chart of a multi-layer perceptron MLP provided in an embodiment of the present application;
[0028] Figure 5 A schematic diagram of the structure of a multi-layer perceptron MLP provided in an embodiment of the present application;
[0029] Figure 6 The second bar chart of indicator weights of a credit decision-making method provided in an embodiment of the present application;
[0030] Figure 7 A prediction flow chart of a multi-layer perceptron MLP provided in an embodiment of the present application;
[0031] Figure 8 One of the structural schematic diagrams of a credit decision-making device provided in an embodiment of the present application;
[0032] Fig. 9 A second structural diagram of a credit decision-making device provided in an embodiment of the present application;
[0033] Fig.10 A schematic diagram of the structure of another credit decision-making device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0034] The credit decision method, device and storage medium provided in the embodiments of the present application are described in detail below in conjunction with the accompanying drawings.
[0035] The term "and / or" in this article is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone.
[0036] The terms "first" and "second" and the like in the specification and drawings of this application are used to distinguish different objects, or to distinguish different processing of the same object, rather than to describe a specific order of objects.
[0037] In addition, the terms "including" and "having" and any variations thereof mentioned in the description of the present application 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 limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or devices.
[0038] It should be noted that, in the embodiments of the present application, words such as "exemplarily" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplarily" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplarily" or "for example" is intended to present related concepts in a specific way.
[0039] In the description of the present application, unless otherwise specified, “plurality” means two or more.
[0040] At present, facing the phenomenon of small scale, low profit and insufficient collateral assets of small and medium-sized enterprises, banks usually make credit decisions based on national policies, enterprise transaction invoices and the influence of upstream and downstream supply and demand. Due to different selected indicators and construction methods, the credit models constructed by various banks are also different.
[0041] In the existing technology, credit decisions for small and medium-sized enterprises are mainly based on the analytic hierarchy process, which is based on the evaluator's understanding of the essence and elements of the evaluation problem. Therefore, it has more qualitative analysis and qualitative judgment. For credit decisions, we often need more quantitative analysis. And the weights of the analytic hierarchy process are often difficult to determine when there are too many indicators and the amount of statistical data is large. Because the results of the above qualitative analysis are often unconvincing; and for enterprises without credit records, most models do not predict and evaluate their risks. This leads to unreasonable lending strategies for banks to enterprises.
[0042] In order to solve the problem of unreasonable lending strategies of banks to different enterprises in the prior art, this application provides a credit decision method, which can use the entropy weight method to calculate the weight of the quantitative index value of the enterprise; then calculate the credit risk value of the enterprise according to the quantitative index value and the weight of the quantitative index value, and then use data mining software to perform linear regression on the credit rating and annual loan interest rate of the enterprise to obtain a first linear programming model, and use the credit risk value calculated above to optimize the first linear programming model to obtain a target linear programming model, and finally input the credit rating of the enterprise that needs a loan into the target linear programming model to obtain the bank's loan strategy for the enterprise. Since the calculation method of the above entropy weight method is a quantitative calculation method, it effectively avoids the influence of the user's subjective qualitative judgment on the weight of each quantitative index value, and ensures the objectivity and accuracy of the weight of each quantitative index value; and in the subsequent process of establishing the model, the credit risk value that can reflect the debt repayment ability of each enterprise is introduced to optimize the model, which improves the prediction accuracy of the model, so that the final determined loan strategy is more in line with the actual debt repayment ability of the enterprise, thereby ensuring the rationality of the bank's loan strategy for different enterprises.
[0043] like Figure 1 As shown, it is a flowchart of the credit decision method provided in an embodiment of the present application, and the method includes the following steps S101 to S105:
[0044] S101. Use the entropy weight method to calculate the weight of the enterprise's quantitative indicator value to obtain the weight of the quantitative indicator value.
[0045] In the embodiment of the present application, the above-mentioned quantitative index value is used to characterize the credit risk situation of the enterprise.
[0046] In the embodiment of the present application, the entropy weight method is an objective value assignment method. In the specific use process, the entropy weight method calculates the entropy weight of each indicator according to the variation degree of each indicator by using information entropy, and then corrects the weight of each indicator by entropy weight, so as to obtain a more objective indicator weight.
[0047] For example, if the entropy weight method of the information entropy indicator weight determination method of a certain indicator is smaller, it means that the variation degree of the indicator value is greater, the amount of information provided is more, the role it can play in the comprehensive evaluation is greater, and its weight is greater. On the contrary, if the entropy weight method of the information entropy indicator weight determination method of a certain indicator is larger, it means that the variation degree of the indicator value is smaller, the amount of information provided is less, the role it plays in the comprehensive evaluation is smaller, and its weight is smaller.
[0048] Example 1: The software code for calculating the weight of the above quantitative index value is as follows:
[0049]
[0050] Exemplarily, the weight of the above quantitative index value is used to characterize the influence of the quantitative index value on the credit risk of the enterprise. The greater the weight, the greater the influence, and the smaller the weight, the smaller the influence.
[0051] In the embodiment of the present application, the above-mentioned quantitative index value is a plurality of index values.
[0052] Exemplarily, one indicator value corresponds to one weight.
[0053] S102. Calculate the credit risk value of the enterprise based on the quantitative indicator value and the weight of the quantitative indicator value.
[0054] In the embodiment of the present application, the credit risk value may be a credit risk score.
[0055] In the embodiment of the present application, after obtaining the weights of the above-mentioned quantitative indicator values, the credit risk score of the enterprise can be calculated according to a preset calculation formula.
[0056] For example, the above quantitative index value is v(i), and the weight of the above quantitative index value is w(i). The credit risk score R can be calculated by formula (1). Formula (1) is as follows:
[0057]
[0058] Among them, i is the label of the quantitative index value.
[0059] Example 2: The software code for calculating the above credit risk value is as follows:
[0060] s = sum(E);
[0061] W1=(1-E) / (ns); % entropy weight method weight
[0062] % Calculate risk score
[0063] For i=1:m
[0064] SS(i)=-W1(1)*x(i,1)+W1(2)*x(i,2)-W1(3)*x(i,3)-W1(4)*x(i,4)+W1
[0065] (5)*x(i,5)-W1(6)*x(i,6)+W1(7)*X(i,7)-W1(8)*x(i,8);
[0066] end
[0067] S103. Using data mining software, linear regression is performed on the credit rating of the enterprise and the annual loan interest rate to obtain a first linear programming model.
[0068] In an embodiment of the present application, the first linear programming model is used to characterize the relationship between the credit rating and the annual interest rate of the loan.
[0069] Exemplarily, the data mining software may be statistical analysis software (Statistical Product and Service Solutions, SPSS).
[0070] In the embodiment of the present application, the credit rating of each enterprise can be obtained by using the k-means clustering algorithm according to the frequency of enterprise default, and then divided into four categories: A, B, C, and D. Then, the credit ratings A, B, C, and D are used as independent variables, and the annual loan interest rate is used as the dependent variable. SPSS software is used for linear regression to obtain the relationship model between the customer churn rate of each credit rating and the annual loan interest rate.
[0071] For example, the estimation results of the relationship model between the customer churn rate of each credit rating and the annual loan interest rate are shown in Table 1 below.
[0072] It should be noted that Table 1 only shows the relationship model between the customer churn rate and the annual loan interest rate of some enterprises. In practical applications, the estimated results of the relationship model of more enterprises may also be included, which will not be repeated here.
[0073] Table 1
[0074]
[0075] For example, let the loan amount of the enterprise be a i The annual interest rate of the above loan is b i , the customer churn rate is c i , the multiple under each credit level is T i , the company's net lending income is E i , then the first linear programming model can be generated according to formula (2). Formula (2) is as follows:
[0076]
[0077]
[0078] Example 3: The specific LINGO solution code for the first linear programming model is as follows:
[0079] Model:
[0080] sets:
[0081] gs / 1..245 / :a,b,c,E,R,T,p;
[0082] endsets
[0083] data:
[0084] T = @ole('C:\Users\Lenovo\Desktop / New1.xlsx);
[0085] R = @ole('C:\Users\Lenovo\Desktop / New 2.xlsx);
[0086] P = @ole('C:\Users\Lenovo\Desktop / New3.xlsx);
[0087] @ole('C:\Users\Lenovo\Desktop / Strategy2.xlsx','A2:A246')=a;
[0088] @ole('C:\Users\Lenovo\Desktop / Strategy2.xlsx','B2:B246')=b;
[0089] enddata
[0090] max=@sum(gs(i):E(i)*(1-c(i)));
[0091] @for(gs(i):E(i)=a(i)*b(i)*(1-R(i))-a(i)*R(i));
[0092] @for(gs(i):c(i)=T(i)*b(i));
[0093] @for(gs(i):a(i)>=10); @for(gs(i):a(i)<=100);
[0094] @for(gs(i):b(i)>=0.04); @for(gs(i):b(i)<=0.15);
[0095] @sum(gs(i):(i))<=10000;
[0096] @for(gs(i):a(i)<=500000*p(i));
[0097] end
[0098] S104. Optimize the first linear programming model according to the credit risk value to obtain a target linear programming model.
[0099] In an embodiment of the present application, the prediction accuracy of the target linear programming model is greater than the prediction accuracy of the first linear programming model.
[0100] In the embodiment of the present application, the above-mentioned first linear programming model has certain shortcomings and does not take the size of the enterprise into consideration.
[0101] For example, under a certain total loan amount of the bank, the same loan amount will be given to enterprises of different sizes, resulting in unreasonable loan strategies for enterprises of different sizes. Therefore, in terms of loan amount allocation, this application introduces the credit risk value to modify the model.
[0102] For example, the data may be screened to exclude enterprises with a credit rating of D, and the credit risk values R of enterprises with credit ratings of A, B, and C may be obtained.
[0103] It should be noted that the above credit risk value can reflect the purchasing power of enterprises. The credit risk value of large enterprises is much higher than that of small enterprises. Therefore, large enterprises have strong purchasing power. Therefore, when the bank's loan limit is limited, the allocation strategy can fully consider the loan needs of large and small enterprises in the market, thereby limiting the loan amount of small enterprises and allocating the amount more reasonably.
[0104] For example, the credit risk value R of each enterprise can be introduced into the first linear programming model, and the loan amount a of each enterprise can be established according to formula (3): i The relationship between the credit risk value of each enterprise. Formula (3) is as follows:
[0105] a i ≤500000p i Formula (3)
[0106] S105. Input the credit rating of the target enterprise into the target linear programming model for prediction to obtain the loan strategy of the target enterprise.
[0107] In an embodiment of the present application, the above-mentioned loan strategy may include loan amount, loan interest rate, etc.
[0108] Illustratively, the loan amount and loan interest rate of the above-mentioned enterprise are shown in Table 2 below.
[0109] It should be noted that Table 2 only shows the loan amounts and loan interest rates of some enterprises. In actual applications, the loan amounts and loan interest rates of more enterprises may also be included, which will not be repeated here.
[0110] Table 2
[0111]
[0112] In the credit decision method provided in the embodiment of the present application, the entropy weight method can be used to calculate the weight of the quantitative index value of the enterprise; then, according to the quantitative index value and the weight of the quantitative index value, the credit risk value of the enterprise is calculated, and then the data mining software is used to perform linear regression on the credit rating and annual loan interest rate of the enterprise to obtain a first linear programming model, and the credit risk value calculated above is used to optimize the first linear programming model to obtain a target linear programming model, and finally the credit rating of the enterprise that needs a loan is input into the target linear programming model to obtain the bank's loan strategy for the enterprise. Since the calculation method of the above entropy weight method is a quantitative calculation method, it effectively avoids the influence of the user's subjective qualitative judgment on the weight of each quantitative index value, and ensures the objectivity and accuracy of the weight of each quantitative index value; and in the subsequent process of establishing the model, the credit risk value that can reflect the debt repayment ability of each enterprise is introduced to optimize the model, which improves the prediction accuracy of the model, so that the final determined loan strategy is more in line with the actual debt repayment ability of the enterprise, thereby ensuring the rationality of the bank's loan strategy for different enterprises.
[0113] Optionally, in an embodiment of the present application, the above-mentioned enterprises include enterprises with credit records and enterprises without credit records.
[0114] Optionally, in an embodiment of the present application, the above-mentioned quantitative index value includes a first quantitative index value and a second quantitative index value.
[0115] Further optionally, in the embodiment of the present application, before the above step S101, the credit decision method provided in the embodiment of the present application may also include the following steps S101a and S101b:
[0116] S101a. When the enterprise has a credit record, obtain a first quantitative index value of the enterprise with a credit record.
[0117] In an embodiment of the present application, the above-mentioned first quantitative indicator value includes: average price tax of input invoices, standard deviation of input invoices, average price tax of output invoices, standard deviation of output invoices, proportion of invalid and negative invoices, frequency of upstream enterprises, frequency of downstream enterprises and credit rating.
[0118] In the embodiment of the present application, the above-mentioned first quantitative index value can reflect the enterprise's strength, stability of supply and demand relationship and reputation.
[0119] For example, the above average price tax of input invoices and average price tax of output invoices can reflect the profitability of an enterprise, and profitability is the ability of an enterprise to make profits in the course of its operations.
[0120] It should be noted that the purpose of operation of small and medium-sized enterprises is to make profits. Sufficient profitability is a necessary basis for repaying debts. The stronger the profitability, the lower the corporate credit risk.
[0121] In addition, the above-mentioned average price tax on input invoices and average price tax on output invoices also reflect the tax payment situation of the enterprise. The better the tax payment situation is, the more it means that the enterprise has sufficient cash flow and good operating conditions, and the lower the credit risk of the enterprise is.
[0122] For example, the above-mentioned standard deviation of input invoices and standard deviation of output invoices can reflect whether the supply and demand relationship of an enterprise is stable. If the supply and output of an enterprise are stable, it means that the operation of the enterprise is stable and the credit risk of the enterprise is low.
[0123] For example, the above-mentioned ratio of voided and negative invoices can reflect the transaction stability of the enterprise, and thus reflect the enterprise's operational capacity. If the ratio of voided and negative invoices is very low, it means that the enterprise's transactions are stable and there are few transaction failures, indicating that the enterprise's operations are stable and the credit risk of the enterprise is low.
[0124] For example, the greater the frequency of the upstream enterprises and the downstream enterprises mentioned above, the wider the enterprise's supply sources and sales channels are, the stronger its ability to resist risks is, and the lower the credit risk of the enterprise is.
[0125] For example, the above credit rating can reflect the company's previous performance in bank credit services. The higher the company's credit rating, the better the company's performance and the lower the company's credit risk.
[0126] For example, after obtaining the first quantitative index value, the following Figure 2 The analytical process shown is used to conduct a quantitative analysis of the company's credit risk.
[0127] For example, the result of weighting the first quantitative index value using the entropy weight method is as follows: Figure 3 shown.
[0128] Among them, the weights of the average price and tax of input invoices and the standard deviation are similar and the largest. Since small and medium-sized enterprises have limited scale and funds compared with large enterprises, the size of the input reflects the comprehensive strength of the enterprise. That is, if the average price and tax of input invoices of a small and medium-sized enterprise is large and the standard deviation is stable, it means that the enterprise has strong capital accumulation and corresponding strong debt repayment ability.
[0129] Secondly, the frequency of upstream and downstream enterprises has a larger weight. The frequency of upstream and downstream enterprises reflects the supply and demand relationship of the enterprise, and a stable supply and demand relationship is the basis for maintaining the operation of the enterprise. If the supply and demand relationship is unstable, it is difficult for the enterprise to survive and develop, and the credit risk of the enterprise is very high.
[0130] Finally, the weights of the average price tax and standard deviation of sales invoices are small. The proportion of voided and negative invoices and credit rating have the smallest weights, because the proportion of voided and negative invoices represents the stability of transactions, which is caused by some external factors affecting the transaction process. Therefore, this indicator has little impact.
[0131] S101b. In the case that the enterprise has no credit record, obtaining a first quantitative index value and a second quantitative index value of the enterprise has no credit record.
[0132] In an embodiment of the present application, the above-mentioned second quantitative indicator values include: the average input amount, the average output amount, the average input tax amount, the average output tax amount, the average input price and tax, and the average output price and tax.
[0133] In an embodiment of the present application, the first quantitative index value is used to predict the credit rating of an enterprise with no credit record.
[0134] In this way, for enterprises with credit records, eight indicators that can reflect the credit risk of the enterprise are obtained, and for enterprises without credit records, six indicators that can predict the credit rating of the enterprise are obtained, ensuring the integrity of the data used in the subsequent model building.
[0135] Optionally, in the embodiment of the present application, before the above step S101b, the credit decision method provided in the embodiment of the present application may further include the following step S101c:
[0136] S101c. In the case that the enterprise has no credit record, obtain the second quantitative index value and credit rating of the target enterprise with a credit record.
[0137] In an embodiment of the present application, the second quantitative index value of the target enterprise with a credit record matches the second quantitative index value of the target enterprise without a credit record.
[0138] Exemplarily, the target enterprise with a credit record and the enterprise without a credit record are enterprises of similar size. In other words, the credit ratings of the target enterprise with a credit record and the enterprise without a credit record are relatively close.
[0139] Further optionally, in the embodiment of the present application, in combination with the above step S101c, the "obtaining the first quantitative index value of the enterprise with no credit record" in the above step S101b may include the following steps S101b1 and S101b2:
[0140] S101b1. Use a preset neural network to learn the correlation between the second quantitative indicator value and the credit rating of the target enterprise with a credit record, and obtain a target neural network model.
[0141] In the embodiment of the present application, the above-mentioned preset neural network can be a multilayer perceptron (MultilayerPerceptron, MLP).
[0142] The following is an explanation of the above MLP:
[0143] Exemplarily, the above-mentioned MLP is also called an artificial neural network (ANN).
[0144] Exemplarily, the above MLP has a three-layer structure, namely an input layer, a hidden layer and an output layer.
[0145] Exemplarily, the input layer may input a large number of indicator values for prediction, first perform supervised learning, and then use the learned neural network for classification prediction.
[0146] Exemplarily, the above hidden layer includes three basic elements: weight, bias and activation function, and the hidden layer is the main part of neural learning.
[0147] It should be noted that the size of the above weight represents the size of the probability; the role of the above bias is to ensure that the output value calculated through the input layer cannot be activated arbitrarily to correctly classify the samples; the above activation function plays the role of nonlinear mapping, which can introduce nonlinear factors into the neural network, so that the neural network can simulate any nonlinear function.
[0148] For example, a common activation function is a Sigmoid function, which can be expressed by formula (3). Formula (3) is as follows:
[0149]
[0150] Exemplarily, the output layer is a multi-category logistic regression result, that is, softmax regression. Therefore, the output of the output layer is softmax(m2x1+b2), where x1 represents the output of the hidden layer f(m1x+b1).
[0151] For example, Figure 4 As shown, after obtaining the second quantitative indicator value and credit rating of the target enterprise with a credit record, the input indicator and output indicator of the enterprise with a credit record can be input into a preset neural network to obtain the predicted credit rating of the target enterprise with a credit record, and then the predicted credit rating is compared with the actual credit rating of the target enterprise with a credit record, and the correlation between the second quantitative indicator and the credit rating is continuously learned to obtain the target neural network model.
[0152] It should be noted that the target neural network model is the trained MLP neural network, which can accurately predict the credit rating of companies with no credit records.
[0153] For example, Figure 5 The figure shows the MLP neural network model after the training. The MLP neural network model is simulated and established by automatically calculating the number of cells. The hidden layer is one layer, the number of neurons is two, the input indicators are six, and the prediction targets are credit rating and whether there is a default.
[0154] In the embodiments of the present application, Figure 6 As shown in the figure, the above MLP neural network can predict the second quantitative indicator value of enterprises without credit records and obtain the importance of each quantitative indicator value. Among them, the average value of input amount and the average value of input price and tax have the greatest impact on neural network learning, and the importance of other indicators is not much different.
[0155] S101b2. Input the second quantitative index value of the enterprise with no credit record into the target neural network model for prediction to obtain the credit rating of the enterprise with no credit record.
[0156] In the embodiment of the present application, since the enterprise without credit record has not obtained a bank loan, the credit rating of the enterprise without credit record cannot be directly obtained. Therefore, it is necessary to use the MLP neural network to predict the credit rating of the enterprise without credit record based on the credit model of the enterprise with credit record.
[0157] For example, Figure 7 As shown, after obtaining the above-mentioned target neural network model, the input indicators and sales indicators of enterprises with no credit records can be input into the target neural network model, and the credit rating of the enterprises with no credit records can be predicted through calculations of the input layer, hidden layer and output layer.
[0158] Exemplarily, the credit rating of an enterprise with no credit record predicted by the above MLP neural network is shown in Table 3 below.
[0159] It should be noted that Table 3 only shows the credit rating prediction results of some enterprises without credit records. In actual applications, more credit rating prediction results of enterprises without credit records may be included, which will not be repeated here.
[0160] Table 3
[0161] 1 Company Code Sales price tax Input value tax N-Credit Level NC-Credit Rating 2 E124 735612.15 55306.787 B 0.615 3 E125 742638.49 56323.234 B 0.62 4 E126 425638.75 250133.44 C 0.535 5 E127 172005.17 1239.859 C 0.414 6 E128 215925.25 2986.758 C 0.363 7 E129 72263.863 14436.202 C 0.388 8 E130 90949.6 24786.095 C 0.357 9 E131 78062.834 19167.101 C 0.373 10 E132 133228.59 13895.423 C 0.376 11 E133 836176.18 25427.787 B 0.645 12 E134 45920.423 2581.268 C 0.394 13 E135 390451.99 9266.23 A 0.69 14 E136 104801.58 102729.4 B 0.517
[0162] In this way, when facing companies with no credit records, the MLP neural network is used to learn the correlation between the second quantitative indicator value and the credit rating of companies with credit records to predict the correlation of companies with no credit records, ensuring that banks can accurately obtain loan strategies for enterprises when facing various types of enterprises.
[0163] Optionally, in the embodiment of the present application, in combination with the above step S101a or step S101b2, the above step S102 may include the following step S102a and step S102b:
[0164] S102a. Eliminate the proportion of voided and negative invoices in the first quantitative index value of the enterprise to obtain a third quantitative index value.
[0165] In the embodiment of the present application, the above-mentioned enterprise may be an enterprise with a credit record or an enterprise without a credit record.
[0166] In the embodiment of the present application, when conducting quantitative analysis of corporate credit risk, firstly, the data of voided invoices and negative invoices are perspectively analyzed using EXCEL to obtain the ratio of voided and negative invoices. Then, the voided invoices and negative invoices are eliminated, and quantitative indicators other than the ratio of voided and negative invoices are determined.
[0167] For example, the above-mentioned average price tax and standard deviation of input items, average price tax and standard deviation of output items, and frequencies of upstream and downstream enterprises can all be obtained by processing the SPSS model.
[0168] S102b. According to the weights of the third quantitative index value and the proportion of voided and negative invoices, the third quantitative index value and the proportion of voided and negative invoices are weighted and summed to obtain the credit risk value of the enterprise.
[0169] In the embodiment of the present application, after using EXCEL to obtain the proportion of invalid and negative invoices, and using the SPSS model to obtain other first quantitative index values, the proportion of invalid and negative invoices and their weights, as well as the other first quantitative index values and their weights, can be respectively input into the above formula (1) for summation calculation, and the calculation result obtained is the credit risk value of the above enterprise.
[0170] In this way, the proportion of invalid and negative invoices is determined through EXCEL data pivoting, which reduces the calculation load of the SPSS model and indirectly improves the overall forecasting efficiency.
[0171] Optionally, in the embodiment of the present application, for enterprises with credit records, the indicators reflecting the credit risk of enterprises can be obtained by consulting the literature, mainly including the profitability, operating capacity, debt-paying capacity, stability of supply and demand relationship, credit rating and other indicators of the enterprise; the eight indicators that can be finally determined to assess the credit risk are: the average price tax of input invoices, the standard deviation of input invoices, the average price tax of output invoices, the standard deviation of output invoices, the proportion of invalid and negative invoices, the frequency of upstream enterprises, the frequency of downstream enterprises, and the credit rating. Then the entropy weight method is used to solve the weight of each indicator, and then the risk score of each enterprise is obtained to establish a risk quantification system. Then, the relationship between the annual loan interest rate and the customer churn rate under different credit ratings is linearly regressed using SPSS software, and the quantitative relationship between the two under different credit ratings is established. Finally, when the total amount of bank loans is fixed, various constraints (such as credit risk scores) are considered, and the maximum bank profit is taken as the objective function to establish a linear optimization model to obtain the final credit strategy allocation model.
[0172] In addition, for enterprises without credit records, neural networks are first used to study enterprises with credit records to obtain the credit rating of each enterprise. Then, based on the credit risk quantification system established above and the credit strategy allocation model based on maximum return, the risk quantification score and credit strategies such as loan amount and loan interest rate of each enterprise are obtained.
[0173] The embodiment of the present application can divide the credit decision-making device into functional modules or functional units according to the above method example. For example, each functional module or functional unit can be divided according to each function, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware or in the form of software functional modules or functional units. Among them, the division of modules or units in the embodiment of the present application is schematic, which is only a logical function division. There may be other division methods in actual implementation.
[0174] like Figure 8, which is a schematic diagram of the structure of a credit decision-making device provided in an embodiment of the present application, the device includes: a calculation unit 201, a processing unit 202 and a prediction unit 203.
[0175] Among them, the above-mentioned calculation unit 201 is used to use the entropy weight method to perform weight calculation on the quantitative indicator value of the enterprise to obtain the weight of the above-mentioned quantitative indicator value, and the above-mentioned quantitative indicator value is used to characterize the credit risk situation of the enterprise; the above-mentioned calculation unit 201 is also used to calculate the credit risk value of the enterprise according to the above-mentioned quantitative indicator value and the weight of the above-mentioned quantitative indicator value; the above-mentioned processing unit 202 is used to use data mining software to perform linear regression on the credit rating and annual loan interest rate of the enterprise to obtain a first linear programming model, and the first linear programming model is used to characterize the relationship between the above-mentioned credit rating and the above-mentioned annual loan interest rate; the above-mentioned processing unit 202 is also used to optimize the above-mentioned first linear programming model according to the above-mentioned credit risk value to obtain a target linear programming model, and the prediction accuracy of the target linear programming model is greater than the prediction accuracy of the above-mentioned first linear programming model; the above-mentioned prediction unit 203 is used to input the credit rating of the target enterprise into the above-mentioned target linear programming model obtained by the processing unit 202 for prediction, and obtain the loan strategy of the above-mentioned target enterprise.
[0176] Optionally, in the embodiment of the present application, the above-mentioned enterprises include enterprises with credit records and enterprises without credit records, and the above-mentioned quantitative index value includes a first quantitative index value and a second quantitative index value, such as Fig. 9 As shown, the above-mentioned device also includes: an acquisition unit 204, wherein: the above-mentioned acquisition unit 204 is used for: before the calculation unit 201 adopts the entropy weight method to perform weight calculation on the quantitative index value of the enterprise, in the case that the above-mentioned enterprise is an enterprise with a credit record, obtaining the first quantitative index value of the enterprise with a credit record, the first quantitative index value includes: the average price and tax of input invoices, the standard deviation of input invoices, the average price and tax of output invoices, the standard deviation of output invoices, the proportion of invalid and negative invoices, the frequency of upstream enterprises, the frequency of downstream enterprises and the credit rating; in the case that the above-mentioned enterprise is an enterprise without a credit record, obtaining the first quantitative index value and the second quantitative index value of the enterprise without a credit record, the second quantitative index value includes: the average value of input amount, the average value of output amount, the average value of input tax, the average value of output tax, the average value of input price and tax and the average value of output price and tax.
[0177] Optionally, in an embodiment of the present application, the acquisition unit 204 is further used to acquire the second quantitative index value and credit rating of the target enterprise with a credit record before acquiring the first quantitative index value of the enterprise with no credit record, in the case that the enterprise is an enterprise with no credit record; wherein the second quantitative index value of the target enterprise with a credit record matches the second quantitative index value of the enterprise with no credit record.
[0178] Optionally, in an embodiment of the present application, the acquisition unit 204 is specifically used to: use a preset neural network to learn the correlation between the second quantitative index value of the target enterprise with a credit record and the credit rating, and obtain a target neural network model; input the second quantitative index value of the enterprise without a credit record into the target neural network model for prediction, and obtain the credit rating of the enterprise without a credit record.
[0179] Optionally, in an embodiment of the present application, the calculation unit 201 is specifically used to: screen out the proportion of the above-mentioned invalidated and negative invoices in the first quantitative index value of the above-mentioned enterprise to obtain a third quantitative index value; and perform weighted summation of the third quantitative index value and the proportion of the above-mentioned invalidated and negative invoices according to the weights of the third quantitative index value and the proportion of the above-mentioned invalidated and negative invoices to obtain the credit risk value of the above-mentioned enterprise.
[0180] In the credit decision-making device provided in the embodiment of the present application, the entropy weight method can be used to calculate the weight of the quantitative index value of the enterprise; then, according to the quantitative index value and the weight of the quantitative index value, the credit risk value of the enterprise is calculated, and then the data mining software is used to perform linear regression on the credit rating and annual loan interest rate of the enterprise to obtain a first linear programming model, and the credit risk value calculated above is used to optimize the first linear programming model to obtain a target linear programming model, and finally the credit rating of the enterprise that needs a loan is input into the target linear programming model to obtain the bank's loan strategy for the enterprise. Since the calculation method of the above entropy weight method is a quantitative calculation method, it effectively avoids the influence of the user's subjective qualitative judgment on the weight of each quantitative index value, and ensures the objectivity and accuracy of the weight of each quantitative index value; and in the subsequent process of establishing the model, the credit risk value that can reflect the debt repayment ability of each enterprise is introduced to optimize the model, which improves the prediction accuracy of the model, so that the final determined loan strategy is more in line with the actual debt repayment ability of the enterprise, thereby ensuring the rationality of the bank's loan strategy for different enterprises.
[0181] Fig.10 Another possible structural diagram of the credit decision device involved in the above embodiment is shown. The credit decision device includes: a processor 302 and a communication interface 303. The processor 302 is used to control and manage the actions of the credit decision device, for example, to execute the steps performed by the above-mentioned calculation unit 201, processing unit 202, prediction unit 203 and acquisition unit 204, and / or to execute other processes of the technology described in this article. The communication interface 303 is used to support the communication between the credit decision device and other network entities. The credit decision device may also include a memory 301 and a bus 304, and the memory 301 is used to store program code and data of the credit decision device.
[0182] Among them, the memory 301 can be a memory in a credit decision-making device, etc. The memory can include a volatile memory, such as a random access memory; the memory can also include a non-volatile memory, such as a read-only memory, a flash memory, a hard disk or a solid-state drive; the memory can also include a combination of the above types of memory.
[0183] The processor 302 may be a processor that implements or executes various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of the present application. The processor may be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field programmable gate array, or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of the present application. The processor may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.
[0184] The bus 304 may be an Extended Industry Standard Architecture (EISA) bus, etc. The bus 304 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.10 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.
[0185] Through the description of the above implementation methods, technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.
[0186] An embodiment of the present application provides a computer program product comprising instructions. When the computer program product is run on a computer, the computer is enabled to execute the credit decision method in the above method embodiment.
[0187] The embodiment of the present application also provides a computer-readable storage medium, in which instructions are stored. When the instructions are executed on a computer, the computer executes the credit decision method in the method flow shown in the above method embodiment.
[0188] Among them, the computer readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a register, a hard disk, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above, or any other form of computer readable storage medium known in the art. An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an application-specific integrated circuit (ASIC). In the embodiments of the present application, a computer-readable storage medium may be any tangible medium that contains or stores a program, which may be used by or in conjunction with an instruction execution system, apparatus, or device.
[0189] An embodiment of the present invention provides a computer program product comprising instructions, which, when executed on a computer, causes the computer to execute the following Figures 1 to 7 The credit decision-making method described in .
[0190] Since the credit decision-making device, computer-readable storage medium, and computer program product in the embodiments of the present invention can be applied to the above-mentioned method, the technical effects that can be obtained can also refer to the above-mentioned method embodiments, and the embodiments of the present invention will not be repeated here.
[0191] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as 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 devices or units, which can be electrical, mechanical or other forms.
[0192] 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 on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0193] 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.
[0194] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto, and any changes or substitutions within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
Claims
1. A credit decision method, characterized in that: The method comprises: The entropy weight method is used to calculate the weight of the quantitative index value of the enterprise to obtain the weight of the quantitative index value, and the quantitative index value is used to characterize the credit risk situation of the enterprise; Calculating the credit risk value of the enterprise according to the quantitative index value and the weight of the quantitative index value; Using data mining software, linear regression is performed on the credit rating and annual loan interest rate of the enterprise to obtain a first linear programming model, wherein the first linear programming model is used to characterize the relationship between the credit rating and the annual loan interest rate; According to the credit risk value, optimizing the first linear programming model to obtain a target linear programming model, wherein the prediction accuracy of the target linear programming model is greater than the prediction accuracy of the first linear programming model; The credit rating of the target enterprise is input into the target linear programming model for prediction to obtain the loan strategy of the target enterprise.
2. The method according to claim 1, characterized in that The enterprises include enterprises with credit records and enterprises without credit records, and the quantitative index value includes a first quantitative index value and a second quantitative index value; Before using the entropy weight method to calculate the weight of the quantitative index value of the enterprise, the method further includes: In the case that the enterprise is an enterprise with a credit record, obtaining a first quantitative index value of the enterprise with a credit record, wherein the first quantitative index value includes: average price tax of input invoices, standard deviation of input invoices, average price tax of output invoices, standard deviation of output invoices, proportion of voided and negative invoices, frequency of upstream enterprises, frequency of downstream enterprises and credit rating; In the case that the enterprise has no credit record, a first quantitative index value and a second quantitative index value of the enterprise with no credit record are obtained, and the second quantitative index value includes: an average value of input amount, an average value of output amount, an average value of input tax, an average value of output tax, an average value of input price and tax, and an average value of output price and tax.
3. The method according to claim 2, characterized in that Before obtaining the first quantitative index value of the enterprise with no credit record, the method further includes: In the case where the enterprise is an enterprise without a credit record, obtaining a second quantitative index value and a credit rating of a target enterprise with a credit record; The second quantitative index value of the target enterprise with a credit record matches the second quantitative index value of the enterprise without a credit record.
4. The method according to claim 3, characterized in that The obtaining of the first quantitative index value of the enterprise with no credit record comprises: Using a preset neural network, learning the correlation between the second quantitative indicator value and the credit rating of the target enterprise with a credit record, and obtaining a target neural network model; The second quantitative index value of the enterprise with no credit record is input into the target neural network model for prediction to obtain the credit rating of the enterprise with no credit record.
5. The method according to claim 2 or 4, characterized in that: The step of calculating the credit risk value of the enterprise according to the quantitative index value and the weight of the quantitative index value includes: The proportion of the voided and negative invoices in the first quantitative index value of the enterprise is screened out to obtain a third quantitative index value; According to the weights of the third quantitative index value and the proportion of the voided and negative invoices, the third quantitative index value and the proportion of the voided and negative invoices are weighted and summed to obtain the credit risk value of the enterprise.
6. A credit decision device, characterized in that: The device comprises: a calculation unit, a processing unit and a prediction unit, wherein: The calculation unit is used to perform weight calculation on the quantitative index value of the enterprise by using the entropy weight method to obtain the weight of the quantitative index value, and the quantitative index value is used to characterize the credit risk situation of the enterprise; The calculation unit is further used to calculate the credit risk value of the enterprise according to the quantitative index value and the weight of the quantitative index value; The processing unit is used to use data mining software to perform linear regression on the credit rating and annual loan interest rate of the enterprise to obtain a first linear programming model, wherein the first linear programming model is used to characterize the relationship between the credit rating and the annual loan interest rate; The processing unit is further used to optimize the first linear programming model according to the credit risk value to obtain a target linear programming model, wherein the prediction accuracy of the target linear programming model is greater than the prediction accuracy of the first linear programming model; The prediction unit is used to input the credit rating of the target enterprise into the target linear programming model obtained by the processing unit for prediction, so as to obtain the loan strategy of the target enterprise.
7. The device according to claim 6, characterized in that The enterprises include enterprises with credit records and enterprises without credit records, the quantitative index value includes a first quantitative index value and a second quantitative index value, and the device further includes: an acquisition unit, wherein: The acquisition unit is used to: Before the calculation unit uses the entropy weight method to perform weight calculation on the quantitative index value of the enterprise, if the enterprise is an enterprise with a credit record, a first quantitative index value of the enterprise with a credit record is obtained, and the first quantitative index value includes: average price tax of input invoices, standard deviation of input invoices, average price tax of output invoices, standard deviation of output invoices, proportion of voided and negative invoices, frequency of upstream enterprises, frequency of downstream enterprises and credit rating; In the case that the enterprise has no credit record, a first quantitative index value and a second quantitative index value of the enterprise with no credit record are obtained, and the second quantitative index value includes: an average value of input amount, an average value of output amount, an average value of input tax, an average value of output tax, an average value of input price and tax, and an average value of output price and tax.
8. The device according to claim 7, characterized in that The acquisition unit is further configured to acquire, before acquiring the first quantitative index value of the enterprise without a credit record, a second quantitative index value and a credit rating of a target enterprise with a credit record if the enterprise is an enterprise without a credit record; The second quantitative index value of the target enterprise with a credit record matches the second quantitative index value of the enterprise without a credit record.
9. The device according to claim 8, characterized in that The acquisition unit is specifically used for: Using a preset neural network, learning the correlation between the second quantitative indicator value and the credit rating of the target enterprise with a credit record, and obtaining a target neural network model; The second quantitative index value of the enterprise with no credit record is input into the target neural network model for prediction to obtain the credit rating of the enterprise with no credit record.
10. The device according to claim 7 or 9, characterized in that The computing unit is specifically used for: The proportion of the voided and negative invoices in the first quantitative index value of the enterprise is screened out to obtain a third quantitative index value; According to the weights of the third quantitative index value and the proportion of the voided and negative invoices, the third quantitative index value and the proportion of the voided and negative invoices are weighted and summed to obtain the credit risk value of the enterprise.
11. A credit decision device, characterized in that: include: A processor and a communication interface; the communication interface is coupled to the processor, and the processor is used to run a computer program or instruction to implement the credit decision method as described in any one of claims 1-5.
12. A computer-readable storage medium, wherein instructions are stored in the computer-readable storage medium, characterized in that: When a computer executes the instruction, the computer executes the credit decision method described in any one of claims 1 to 5.
