A method, device, equipment and medium for generating a credit risk assessment model

Abstract

The present invention discloses a method, device, equipment and medium for generating a credit risk assessment model. The method comprises: constructing assessment indicators based on data from governments at all levels, urban investment entity data and public opinion data; the assessment indicators include quantitative indicators and qualitative indicators; the qualitative indicators at least include public opinion risk labels; the public opinion risk labels are obtained by natural language processing of public opinion data; determining a data conversion model based on the mean and standard deviation of the quantitative indicators in the sample data; converting the quantitative indicators into risk factors based on the data conversion model to obtain a first feature; selecting a target feature from the first feature and the qualitative indicator based on the correlation analysis results of the first feature and the correlation analysis results of the qualitative indicator; constructing a target credit risk assessment model based on the target feature, and determining the parameters of the target credit risk assessment model based on the sample data. The embodiment of the present invention can accurately assess the credit risk level of the urban investment entity.

Description

Technical Field

[0001] The present invention relates to the technical field of credit risk assessment, and in particular to a method, device, equipment and medium for generating a credit risk assessment model. Background Art

[0002] Before investing in municipal bonds, it is necessary to assess the credit default risk level of the municipal investment entity, estimate the risk exposure, and do a good job of risk management. Summary of the invention

[0003] The present invention provides a method, device, equipment and medium for generating a credit risk assessment model to solve the problem that investors cannot clearly understand the credit status of urban investment entities and cannot reduce investment risks.

[0004] According to one aspect of the present invention, there is provided a method for generating a credit risk assessment model, comprising:

[0005] Constructing evaluation indicators based on data from governments at all levels, data from urban investment entities, and public opinion data; the evaluation indicators include quantitative indicators and qualitative indicators; the qualitative indicators at least include public opinion risk labels; the public opinion risk labels are obtained by performing natural language processing on the public opinion data;

[0006] Determine a data conversion model based on the average value and standard deviation of the quantitative indicator in the sample data; perform risk factor conversion on the quantitative indicator based on the data conversion model to obtain a first feature;

[0007] Selecting a target feature from the first feature and the qualitative indicator according to a correlation analysis result of the first feature and a correlation analysis result of the qualitative indicator;

[0008] Constructing a target credit risk assessment model according to the target characteristics, and determining parameters of the target credit risk assessment model based on sample data;

[0009] The process of determining the public opinion risk label includes:

[0010] Perform entity recognition and relationship extraction on the initial news data to obtain the target news data associated with the target municipal investment entity;

[0011] Performing text classification on the target news data to determine the target risk type of the target news data;

[0012] Determine the candidate risk tags included in the target risk type based on the association relationship between the preset risk types and the risk tags;

[0013] According to the text similarity between the target news data and the historical news data associated with the candidate risk label, the target risk label of the target news data is determined as the public opinion risk label of the target urban investment entity.

[0014] According to another aspect of the present invention, there is provided a credit risk assessment model generating device, comprising:

[0015] An indicator construction module is used to construct evaluation indicators based on data from governments at all levels, urban investment entities and public opinion data; the evaluation indicators include quantitative indicators and qualitative indicators; the qualitative indicators at least include public opinion risk labels; the public opinion risk labels are obtained by performing natural language processing on the public opinion data; wherein the process of determining the public opinion risk labels includes: performing entity recognition and relationship extraction on the initial news data to obtain target news data associated with the target urban investment entity; performing text classification on the target news data to determine the target risk type of the target news data; based on the association relationship between the pre-set risk type and the risk label, determining the candidate risk labels included in the target risk type; and determining the target risk label of the target news data as the public opinion risk label of the target urban investment entity based on the text similarity between the target news data and the historical news data associated with the candidate risk label.

[0016] An indicator conversion module, used to determine a data conversion model based on the average value and standard deviation of the quantitative indicator in the sample data; and to convert the quantitative indicator into a risk factor based on the data conversion model to obtain a first feature;

[0017] A feature screening module, used for screening target features from the first features and the qualitative indicators according to the correlation analysis results of the first features and the correlation analysis results of the qualitative indicators;

[0018] The assessment model determination module is used to construct a target credit risk assessment model according to the target characteristics and determine the parameters of the target credit risk assessment model based on sample data.

[0019] According to another aspect of the present invention, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the method for generating a credit risk assessment model according to any embodiment of the present invention is implemented.

[0020] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0021] At least one processor; and a memory in communication with the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the credit risk assessment model generation method described in any embodiment of the present invention.

[0022] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the credit risk assessment model generation method described in any embodiment of the present invention when executed.

[0023] The embodiment of the present invention constructs evaluation indicators of governments at all levels and evaluation indicators of urban investment entities by collecting and processing the public financial report data of listed companies and bond issuers associated with urban investment entities, the public information of enterprises associated with urban investment entities, and the public economic data, regulatory announcements, and news data of governments at all levels in the country and the region where the urban investment entities are located. Based on the data of the urban investment entities themselves, the impact of governments at all levels and public opinion on the credit risk of urban investment entities is fully considered, and the financial status, operating ability, project quality and other aspects of the urban investment entities can be fully reflected. Through the evaluation indicators, investors can have a clearer understanding of the credit status of urban investment entities, so as to make more rational investment decisions and reduce investment risks. By constructing public opinion risk labels, the market public opinion information of urban investment entities can be monitored and analyzed in real time. Through public opinion risk labels, investors can have a more comprehensive understanding of the market image and reputation of urban investment entities, thereby enhancing investment confidence. The first credit risk assessment model and the second credit risk assessment model are combined to construct a target credit risk assessment model, so that the constructed target credit risk assessment model can more accurately assess the credit risk level of the urban investment entity.

[0024] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0026] Figure 1 is a first flow chart of a method for generating a credit risk assessment model provided by an embodiment of the present invention;

[0027] Figure 2 is a second flow chart of the credit risk assessment model generation method provided by an embodiment of the present invention;

[0028] Figure 3 is a structural schematic diagram of a credit risk assessment model generating device provided in an embodiment of the present invention;

[0029] Figure 4 It is a schematic diagram of the structure of an electronic device implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0031] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] Figure 1 This is the first flow chart of the credit risk assessment model generation method provided by the embodiment of the present invention. This embodiment can be applied to the case where a credit risk assessment model is constructed for an urban investment entity to conduct credit risk assessment on the urban investment entity. The method can be executed by a credit risk assessment model generation device, which can be implemented in the form of hardware and / or software, and the device can be configured in an electronic device with corresponding data processing capabilities. Figure 1 As shown, the method includes:

[0033] S110. Construct evaluation indicators based on data from governments at all levels, data from urban investment entities and public opinion data.

[0034] The evaluation indicators include quantitative indicators and qualitative indicators; the qualitative indicators at least include public opinion risk labels; the public opinion risk labels are obtained by performing natural language processing on the public opinion data.

[0035] Urban investment entities, namely urban investment companies, are short for urban investment companies, also known as "local government financing platforms". These companies are mainly established by local governments and their departments and institutions through fiscal appropriations or injection of land, equity and other assets, and have independent legal personality. The core function of urban investment companies is to raise funds for local governments. These funds are mainly used for the construction of public welfare or quasi-public welfare projects in areas such as urban infrastructure, such as urban roads and bridges, water supply and heating, rail transit, and affordable housing projects. Specifically, collect public financial report data disclosed by listed companies and bond issuers associated with urban investment entities in exchanges, inter-bank trading markets, etc.; collect information on enterprises associated with urban investment entities disclosed in the national enterprise credit information disclosure system. Collect economic data disclosed by governments at all levels in the country and the regions where urban investment entities are located. Collect announcements disclosed by regulatory agencies in real time.

[0036] Data collection requires specifying a collection period to determine whether the municipal investment entity has defaulted during the collection period. For any collection period, set the first deadline and the second deadline. The first deadline is earlier than the second deadline, and the two deadlines are at least 12 months apart. For the pre-set municipal investment entity, collect relevant data of the municipal investment entity within a certain period before the first deadline, including: public financial report data of listed companies and bond issuers associated with the municipal investment entity, public information of enterprises associated with the municipal investment entity, economic data disclosed by governments at all levels in the country and the region where the municipal investment entity is located, regulatory announcements and news data, etc. Add the relevant data of the municipal investment entity within a certain period before the first deadline to the current collection period of the municipal investment entity as the collection data of the current collection period of the municipal investment entity.

[0037] For example, considering that the lag period of financial data is 3 months, the first deadline is usually set at the end of March each year. Assuming that the first deadline is April 1, 2024 and the second deadline is March 31, 2025, the financial statements of the relevant enterprises of the municipal investment entity in December 2023 should be added to the current collection cycle. Whether the municipal investment entity defaults during the current collection period from April 1, 2024 to March 31, 2025 is related to the financial statement data of the relevant enterprises of the municipal investment entity in December 2023.

[0038] The collected structured and unstructured data are processed to obtain data from governments at all levels, urban investment entities and public opinion data to construct evaluation indicators for provincial governments, municipal governments, district and county governments, and urban investment entities. The evaluation indicators for urban investment entities include the basic evaluation indicators for urban investment entities and the associated evaluation indicators between local governments and urban investment entities. Local governments refer to the governments with direct jurisdiction over urban investment entities, which can be provincial governments, municipal governments or district and county governments.

[0039] Specifically, through natural language processing technology, unstructured document information is converted into structured data to obtain the financial items and data of the balance sheet, income and expenditure statement, cash flow statement, and notes statement of the urban investment entities and governments at all levels. Based on the data of governments at all levels and urban investment entities, the evaluation indicators of governments at all levels, the basic evaluation indicators of urban investment entities, and the related evaluation indicators of local governments and urban investment entities are constructed.

[0040] For example, the evaluation indicators of provincial governments can be divided into different indicator types, including: economic strength type, fiscal strength type, fiscal flexibility type, debt burden type, etc. The evaluation indicators of municipal governments can be divided into different indicator types, including: economic strength type, fiscal strength type, fiscal flexibility type, debt repayment capacity type, etc. The evaluation indicators of district and county governments can be divided into different indicator types, including: economic strength type, fiscal strength type, fiscal flexibility type, etc. Each indicator type includes specific evaluation indicators, and the evaluation indicators of governments at all levels are shown in Table 1.

[0041] The basic evaluation indicators of urban investment entities can be divided into different types, including solvency type, scale type, growth type, liquidity type, leverage type, profitability type, turnover type and external guarantee type, etc. Each indicator type includes specific evaluation indicators. The basic evaluation indicators of urban investment entities are shown in Table 2.

[0042] Table 1: Evaluation indicators for various levels of government

[0043]

[0044] Table 2: Basic evaluation indicators of urban investment entities

[0045]

[0046] The evaluation indicators of the relationship between local governments and urban investment entities can be divided into: the type of importance of urban investment entities to the government and the type of relationship between the government and urban investment entities. The evaluation indicators of the relationship between local governments and urban investment entities are shown in Table 3.

[0047] Table 3: Evaluation indicators of the relationship between local governments and urban investment entities

[0048]

[0049] The evaluation indicators of the urban investment entity also include the public opinion risk label, which belongs to the public opinion risk type. The public opinion data related to the urban investment entity is obtained from the news data, and the public opinion data is processed by natural language to obtain the public opinion risk label of the urban investment entity.

[0050] Evaluation indicators can be divided into quantitative indicators and qualitative indicators according to whether they can be quantified. Among them, the public opinion risk label is a qualitative indicator.

[0051] In an optional embodiment, the process of determining the public opinion risk label includes: performing entity recognition and relationship extraction on the initial news data to obtain target news data associated with the target urban investment entity; performing text classification on the target news data to determine the target risk type of the target news data; based on the association relationship between the pre-set risk type and the risk label, determining the candidate risk labels included in the target risk type; and determining the target risk label of the target news data as the public opinion risk label of the target urban investment entity based on the text similarity between the target news data and the historical news data associated with the candidate risk labels.

[0052] Specifically, based on the public business data of enterprises, a knowledge graph of enterprises is constructed; entity recognition and relationship extraction are performed on the initial news data, and the target news data associated with the target urban investment entity is obtained in combination with the knowledge graph of the enterprise. Text classification is performed on the target news data to determine the target risk type of the target news data. The association relationship between each risk type and the risk label is preset, and each risk type includes at least one risk label, and the risk label is associated with the historical news data. Based on the association relationship between the pre-set risk type and the risk label, the candidate risk labels included in the target risk type are determined, and then the historical news data associated with the candidate risk labels are determined. The text similarity between the target news data and the historical news data associated with each candidate risk label is determined according to a text similarity algorithm, such as a similarity calculation method based on word vectors or a text matching model based on neural networks. The candidate risk label corresponding to the historical news data with the highest text similarity to the target news data is selected as the target risk label of the target news data, and the target risk label is the public opinion risk label of the target urban investment entity.

[0053] Exemplary public opinion risk labels include: broken capital chain, asset disposal, asset freeze, debt default, high debt ratio, capital shortage, high-interest financing, judicial enforcement, penalties by financial regulatory agencies, administrative investigations, litigation disputes, information disclosure violations, penalties by other departments, regulatory inquiries, arbitration cases, labor disputes, dishonest debtors, debt crisis, bond default, contract default, bad debts, credit limit reduction, etc.

[0054] By collecting and processing the public financial report data of listed companies and bond issuers associated with urban investment entities, the public information of enterprises associated with urban investment entities, the public economic data, regulatory announcements and news data of governments at all levels in the country and the regions where urban investment entities are located, we can build evaluation indicators for governments at all levels and urban investment entities. Based on the data of urban investment entities themselves, we can fully consider the impact of governments at all levels and public opinion on the credit risk of urban investment entities, and comprehensively reflect the financial status, operating capabilities, project quality and other aspects of urban investment entities. By building evaluation indicators, investors can have a clearer understanding of the credit status of urban investment entities, so as to make more rational investment decisions and reduce investment risks. By building public opinion risk tags, we can monitor and analyze the market public opinion information of urban investment entities in real time. Through public opinion risk tags, investors can have a more comprehensive understanding of the market image and reputation of urban investment entities, thereby enhancing investment confidence.

[0055] S120. Determine a data conversion model based on the average value and standard deviation of the quantitative indicator in the sample data; and convert the quantitative indicator into a risk factor based on the data conversion model to obtain a first feature.

[0056] Different quantitative indicators have different data magnitudes. Directly using quantitative indicators of different data magnitudes for credit risk assessment may cause the influence of quantitative indicators with large magnitudes to be over-magnified in the credit risk assessment process, thereby affecting the accuracy of the credit risk assessment results. Moreover, each quantitative indicator and the credit risk assessment result may not necessarily satisfy a linear correlation. Therefore, a data conversion model can be determined based on the mean and standard deviation of the quantitative indicators in the sample data. Based on the data conversion model, the quantitative indicators are converted into risk factors to obtain the first feature, so that different quantitative indicators are converted to the same data magnitude through the data conversion model, and the first feature obtained after the conversion satisfies the linear correlation relationship with the credit risk assessment result.

[0057] In an optional embodiment, for any quantitative indicator, determining the data conversion model based on the mean and standard deviation of the quantitative indicator in the sample data includes: determining a first reference value based on a preset first confidence level and the mean and standard deviation of the quantitative indicator; determining a second reference value based on a preset second confidence level and the mean and standard deviation of the quantitative indicator; determining the data conversion model based on the first confidence level, the first reference value, the second confidence level and the second reference value.

[0058] Optionally, the data conversion model is Logit conversion, and conversion parameters are constructed based on the Sigmiod function as the data conversion model. The specific formula is shown in Formula 1 below:

[0059] (1)

[0060] in, is the value of the quantitative indicator after risk factor conversion; X is the original value of the quantitative indicator; a and b are the conversion parameters of the data conversion model.

[0061] For any quantitative indicator, the conversion parameters of the data conversion model are determined according to the mean value and standard deviation of the quantitative indicator in the sample data. Specifically, the mean value and standard deviation of the quantitative indicator in the sample data are determined, the first confidence level is preset, and the Z value corresponding to the first confidence level under the standard normal distribution is determined. The product of the standard deviation of the quantitative indicator and the Z value corresponding to the first confidence level under the standard normal distribution is determined as the first product, and the mean value of the quantitative indicator minus the first product is used as the first reference value. The details are shown in Formula 2 below.

[0062] (2)

[0063] in, is the first confidence level, is the first reference value, is the average value of the quantitative index, is the standard deviation of the quantitative indicator, is the Z value corresponding to the first confidence level under the standard normal distribution.

[0064] A second confidence level is preset, and the sum of the second confidence level and the first confidence level is 1; the mean value and standard deviation of the quantitative indicator in the sample data are determined, and the Z value corresponding to the second confidence level under the standard normal distribution is determined. The product of the standard deviation of the quantitative indicator and the Z value corresponding to the second confidence level under the standard normal distribution is determined as the second product, and the sum of the mean value of the quantitative indicator and the second product is taken as the second reference value. The details are shown in Formula 3 below.

[0065] (3)

[0066] in, is the second confidence level, is the second reference value, is the average value of the quantitative index, is the standard deviation of the quantitative indicator, is the Z value corresponding to the second confidence level under the standard normal distribution.

[0067] The first confidence level is used as the value after the first reference value is converted by the data conversion model; the second confidence level is used as the value after the second reference value is converted by the data conversion model; and the conversion parameters of the data conversion model are determined based on the first confidence level, the first reference value, the second confidence level and the second reference value.

[0068] By converting the quantitative indicators into risk factors through the data conversion model, the data magnitudes among the quantitative indicators are effectively unified, and the original distribution of the quantitative indicators is "stretched", making the quantitative indicators more capable of distinguishing credit risks. The first feature obtained after the conversion is linearly correlated with the evaluation result, and the converted evaluation indicators are normal. The credit risk of the urban investment entity can be evaluated using multiple linear regression. The credit risk of the urban investment entity can be determined without building a complex credit risk assessment model, thereby improving the efficiency of determining the target credit risk assessment model.

[0069] S130: Filter a target feature from the first feature and the qualitative indicator according to a correlation analysis result of the first feature and a correlation analysis result of the qualitative indicator.

[0070] The target characteristics of governments at all levels shall be determined according to their evaluation indicators; the target characteristics of urban investment entities shall be determined according to their evaluation indicators.

[0071] Specifically, determine the governments at all levels associated with the urban investment entity, conduct a correlation analysis on the first feature converted from the quantitative indicators of any level of government, and select the first target feature from the first feature based on the correlation analysis results of the first feature; and conduct a correlation analysis on the qualitative indicators of any level of government, and select the second target feature from the qualitative indicators based on the correlation analysis results of the qualitative indicators; the first target feature and the second target feature are used as the target characteristics of the current government.

[0072] A correlation analysis is performed on the first feature converted from the quantitative indicators in the evaluation indicators of the urban investment entity, and a first target feature is screened from the first feature based on the correlation analysis result of the first feature; and a correlation analysis is performed on the qualitative indicators in the evaluation indicators of the urban investment entity, and a second target feature is screened from the qualitative indicators based on the correlation analysis result of the qualitative indicators; the first target feature and the second target feature are used as the target features of the urban investment entity.

[0073] By analyzing the results based on the correlation between features or indicators, the features or indicators with high correlation are eliminated, thus avoiding the multicollinearity problem of the multi-factor regression model and making the credit risk assessment results of the target credit risk assessment model more accurate.

[0074] S140: construct a target credit risk assessment model according to the target characteristics, and determine parameters of the target credit risk assessment model based on sample data.

[0075] According to the target characteristics, a target credit risk assessment model can be constructed. A multivariate linear regression model can be selected as the target credit risk assessment model, and the parameters of the target credit risk assessment model can be determined based on the sample data. Optionally, a first model can be constructed based on the target characteristics of governments at all levels, a second model can be constructed based on the target characteristics of the urban investment entity, and a target credit risk assessment model can be constructed by combining the first model and the second model based on the relationship between the urban investment entity and governments at all levels.

[0076] In an optional implementation, constructing a target credit risk assessment model according to the target characteristics and determining the parameters of the target credit risk assessment model based on sample data includes: constructing a first credit risk assessment model according to the target characteristics of governments at all levels; constructing a second credit risk assessment model according to the target characteristics of urban investment entities; and determining the target credit risk assessment model based on the first credit risk assessment model and the second credit risk assessment model.

[0077] As the bond issuance platform of local governments, the financial strength of local governments and the support of local governments for urban investment entities all affect the credit risk assessment results of urban investment entities. Therefore, when assessing the credit default risk of urban investment entities, it is necessary to consider the rating of local governments and use the local government rating as part of the credit risk assessment of urban investment entities.

[0078] Since the support from higher-level governments in finance, policy and other aspects will affect the development of subordinate regions, and the comprehensive strength of higher-level governments is different, the degree of support for subordinate regions is also different. Therefore, the credit risk assessment model of municipal governments and the credit risk assessment model of district and county governments include not only the credit risk assessment results of the government at the same level on urban investment entities, but also the credit risk assessment results of higher-level governments on urban investment entities.

[0079] The first credit risk assessment model is constructed according to the target characteristics of governments at all levels, including: determining the level of local government where the urban investment subject is located. If the local government where the urban investment subject is located is a provincial government, a multivariate linear regression model is constructed as the first credit risk assessment model based on the sample data and target characteristics of the provincial government. If the local government where the urban investment subject is located is a municipal government or a district or county government, a multivariate linear regression model is constructed as the first reference model based on the sample data and target characteristics of the superior government of the local government; a multivariate linear regression model is constructed as the second reference model based on the sample data and target characteristics of the local government where the urban investment subject is located, as shown in the following formula 4; reasonable weights are set for the first reference model and the second reference model according to the degree of influence of the superior government and the government at this level on the credit of the urban investment subject; the first reference model and the second reference model are weighted and summed according to the set weights to obtain the first credit risk assessment model, as shown in the following formula 5.

[0080] (4)

[0081] (5)

[0082] in, Indicates the target characteristics of the government at this level, are the parameters of the multiple linear regression model, It indicates the initial credit risk assessment result of the local government on the urban investment entity. It indicates the final credit risk assessment result of the higher-level government on the urban investment entity. It indicates the final credit risk assessment result of the local government on the urban investment entity. Indicates the influence weight of the higher-level government on the credit of the urban investment entity. When the local government where the urban investment entity is located is a provincial government, Take 0; when the local government where the urban investment entity is located is a municipal government or a district or county government The default value is 0.5. SPSS software can be used to solve the parameters in the first credit risk assessment model, that is, the parameters of the multiple linear regression model.

[0083] A multivariate linear regression model is constructed as the second credit risk assessment model based on the sample data and target characteristics of the urban investment entity. The target credit risk assessment model is determined based on the first credit risk assessment model and the second credit risk assessment model, including setting reasonable weights for the first credit risk assessment model and the second credit risk assessment model based on the degree of influence of the local government where the urban investment entity is located on the credit of the urban investment entity; the first credit risk assessment model and the second credit risk assessment model are weighted and summed according to the set weights to obtain the target credit risk assessment model, as shown in the following formula 6.

[0084] (6)

[0085] in, It indicates the degree of influence of the local government where the urban investment entity is located on the credit of the urban investment entity, and the default value is 0.5; The initial credit risk assessment result of the urban investment entity obtained based on the second credit risk assessment model; It is the final credit risk assessment result of the municipal investment entity.

[0086] By constructing a target credit risk assessment model based on target characteristics and determining the parameters of the target credit risk assessment model based on sample data, the initial urban investment credit risk assessment results are obtained based on the urban investment subject's own data. On the basis of the initial urban investment credit risk assessment results, the impact of governments at all levels on the credit risk of urban investment subjects is fully considered, and the first credit risk assessment model and the second credit risk assessment model are combined to construct a target credit risk assessment model, so that the constructed target credit risk assessment model can more accurately assess the credit risk level of urban investment subjects.

[0087] The embodiment of the present invention constructs evaluation indicators of governments at all levels and evaluation indicators of urban investment entities by collecting and processing the public financial report data of listed companies and bond issuers associated with urban investment entities, the public information of enterprises associated with urban investment entities, and the public economic data, regulatory announcements, and news data of governments at all levels in the country and the region where the urban investment entities are located. Based on the data of the urban investment entities themselves, the impact of governments at all levels and public opinion on the credit risk of urban investment entities is fully considered, and the financial status, operating ability, project quality and other aspects of the urban investment entities can be fully reflected. Through the evaluation indicators, investors can have a clearer understanding of the credit status of urban investment entities, so as to make more rational investment decisions and reduce investment risks. By constructing public opinion risk labels, the market public opinion information of urban investment entities can be monitored and analyzed in real time. Through public opinion risk labels, investors can have a more comprehensive understanding of the market image and reputation of urban investment entities, thereby enhancing investment confidence. The first credit risk assessment model and the second credit risk assessment model are combined to construct a target credit risk assessment model, so that the constructed target credit risk assessment model can more accurately assess the credit risk level of the urban investment entity.

[0088] Figure 2 This is a second flow chart of the credit risk assessment model generation method provided by an embodiment of the present invention. This embodiment is optimized and improved on the basis of the above embodiment. Figure 2 As shown, the method includes:

[0089] S210. Construct evaluation indicators based on data from governments at all levels, data from urban investment entities and public opinion data.

[0090] The evaluation indicators include quantitative indicators and qualitative indicators; the qualitative indicators at least include public opinion risk labels; the public opinion risk labels are obtained by performing natural language processing on the public opinion data.

[0091] S220. Determine a data conversion model based on the mean value and standard deviation of the quantitative indicator in the sample data; and convert the quantitative indicator into a risk factor based on the data conversion model to obtain a first feature.

[0092] S230: If the Pearson correlation coefficient between any group of features in the first feature is greater than or equal to the correlation threshold, select a feature from the group of features as a candidate feature.

[0093] The regression model cannot contain features that are highly correlated with each other. It is necessary to determine which features in the first feature are highly correlated and select one feature from a group of highly correlated features to enter the final credit risk assessment model to ensure the explanatory effect of the regression model and solve the problem of multicollinearity.

[0094] Optionally, the Pearson correlation coefficient is used to determine the similarity between any two first features, and a correlation matrix is ​​established according to the Pearson correlation coefficient R between any two first features to perform correlation analysis between the first features. The calculation formula of the Pearson correlation coefficient is shown in the following formula 7:

[0095] (7)

[0096] in represents the value of the i-th sample in the first feature X, represents the value of the i-th sample in the first feature Y, represents the mean of the sample in the first feature X, represents the mean of the sample in the first feature Y, and R is the correlation coefficient between the first feature X and the first feature Y.

[0097] For example, 0≤|R|<0.3 indicates that X and Y have a low linear correlation; 0.3≤|R|<0.5 indicates that X and Y have a low to medium linear correlation; 0.5≤|R|<0.8 indicates that X and Y have a medium linear correlation; 0.8≤|R|≤1 indicates that X and Y have a high linear correlation. The correlation threshold is preset to 0.8. When the Pearson correlation coefficient between any group of features in the first feature is greater than or equal to the correlation threshold, one feature is selected from the group of features as a candidate feature to enter the final credit risk assessment model.

[0098] In an optional implementation, if the Pearson correlation coefficient between any group of features in the first feature is greater than or equal to a correlation threshold, then selecting a feature from the group of features as a candidate feature includes: determining the degree of influence of any feature in the group of features on the credit score, and selecting the candidate feature from the group of features according to the degree of influence.

[0099] Specifically, in the first features obtained after the quantitative indicators are converted into risk factors, any first feature and the credit score have satisfied a linear relationship, and when the Pearson correlation coefficient between any group of features in the first features is greater than or equal to the correlation threshold, the least square method is used to calculate the influence of each first feature in the group of features on the credit score, including: according to the least square method, the coefficient of the univariate linear regression equation is determined based on the sample data, and the influence of the current first feature on the credit score is determined according to the coefficient of the univariate linear regression equation. The larger the absolute value of the coefficient of the univariate linear regression equation, the greater the influence of the current first feature on the credit score. The details are shown in the following formulas 8.1-8.3. The first feature with the greatest influence on the credit score is selected from the group of features as the candidate feature.

[0100] (8.1)

[0101] (8.2)

[0102] (8.3)

[0103] in represents the value of the i-th sample in the first feature X, represents the credit score of the i-th sample, represents the mean of the sample in the first feature X, Represents the mean credit score of the sample. The larger the absolute value of , the greater the impact of the first feature X on the credit score.

[0104] S240. The importance of discrete values ​​in the qualitative indicators is used as the rank of the sample data under the qualitative indicators.

[0105] Each qualitative indicator includes at least one discrete value. The importance of each discrete value is determined based on expert experience, and an importance value is set for each discrete value under any qualitative indicator. According to the importance value of each discrete value under the qualitative indicator, the rank of the sample data under the qualitative indicator is determined.

[0106] For example, the qualitative indicators of the municipal investment entity include a public opinion risk label, which may include discrete values ​​such as asset disposal, information disclosure violations, regulatory inquiries, and labor disputes. Based on expert experience, set importance values ​​for discrete values ​​under the public opinion risk label, such as 1 for asset disposal, 2 for information disclosure violations, 3 for regulatory inquiries, and 4 for labor disputes. Use the importance value as the rank of each sample under the current qualitative indicator of the sample data, i.e., the public opinion risk label. Based on the same process, determine the rank of the sample data under other qualitative indicators.

[0107] S250. If the sample data have the same rank under the qualitative indicators, the correlation between the qualitative indicators is determined according to the Pearson correlation coefficient; if the sample data have different ranks under any two qualitative indicators, the correlation between the two qualitative indicators is determined according to the Spearman correlation coefficient.

[0108] The sample data includes the value of any evaluation indicator, the credit score of the municipal investment entity, and whether the municipal investment entity has defaulted. Before calculating the correlation between qualitative indicators, the qualitative indicators are converted into WOE to determine the numerical value corresponding to each discrete value under the qualitative indicators. The discrete values ​​under the qualitative indicators in the sample data are converted into numerical values ​​to determine the correlation between the qualitative indicators. The WOE (Weight of Evidence) conversion process is shown in the following formula 9.

[0109] (9)

[0110] Among them, g i b represents the number of samples of municipal investment entities that have not defaulted when the qualitative indicator takes the discrete value i; i It represents the number of samples of default by the municipal investment entity when the qualitative indicator takes the discrete value i; G represents the total number of samples of non-default by the municipal investment entity; B represents the total number of samples of default by the municipal investment entity.

[0111] Determine whether the ranks of the sample data are the same under any qualitative indicator. If the same values ​​exist in the sample data under any qualitative indicator, the sample data have the same rank under the qualitative indicator, and the Pearson correlation coefficient is used to calculate the correlation between qualitative indicators with the same rank; if the values ​​of the sample data under any qualitative indicator are different, the ranks of the sample data under the qualitative indicator are different, and the Spearman correlation coefficient is used to calculate the correlation between qualitative indicators with different ranks. The determination process of the Spearman correlation coefficient is shown in Formula 10.

[0112] (10)

[0113] Among them, d a is the difference between the ranks of sample a under any two qualitative indicators. For example, the rank of sample a under qualitative indicator X is x a , the rank under the qualitative index Y is y a , then the rank difference of sample a under qualitative indicators X and Y is ; n is the number of samples.

[0114] The values ​​of the Spearman correlation coefficient and the Pearson correlation coefficient are both between -1 and 1. When the correlation coefficient is positive, it means that there is a positive correlation between the two indicators; when the correlation coefficient is negative, it means that there is a negative correlation between the two indicators; when the correlation coefficient is 0, it means that there is no linear relationship between the two indicators. Generally speaking, if the absolute value of the correlation coefficient is greater than 0.5, it can be considered that there is a strong correlation between the two indicators; if the absolute value of the correlation coefficient is less than 0.3, it can be considered that there is no obvious correlation between the two indicators.

[0115] S260: If the correlation coefficient between any group of indicators in the qualitative indicators is greater than or equal to the correlation threshold, a qualitative indicator is selected from the group of indicators as a candidate feature.

[0116] Optionally, the degree of influence of any feature in the group of features on the credit score is determined, and a feature to be selected is selected from the group of features according to the degree of influence.

[0117] S270: Use the candidate features and the features whose correlation coefficients are less than a correlation threshold as the target features.

[0118] For example, the target characteristics of governments at all levels are shown in Table 4. The target characteristics of urban investment entities are shown in Table 5.

[0119] Table 4: Characteristics of goals at all levels of government

[0120]

[0121] Table 5: Target characteristics of urban investment entities

[0122]

[0123] By eliminating the highly correlated features or indicators based on the correlation analysis results between the first feature and the qualitative indicators, the multicollinearity problem of the multi-factor regression model is avoided, making the credit risk assessment results of the target credit risk assessment model more accurate.

[0124] S280: construct a target credit risk assessment model according to the target characteristics, and determine parameters of the target credit risk assessment model based on sample data.

[0125] The sample data can be divided into a training set and a validation set, and the parameters of the target credit risk assessment model can be determined based on the training set sample data. Based on the validation set sample data, it is verified whether the credit score of the sample data obtained according to the target credit risk assessment model conforms to the economic assumptions and the actual market results, and the model parameters are optimized according to the verification results.

[0126] The credit score of the urban investment entity is obtained through the target credit risk assessment model, and the credit risk level of the urban investment entity is determined based on the preset mapping relationship between the credit score and the credit risk level.

[0127] The embodiment of the present invention constructs evaluation indicators of governments at all levels and evaluation indicators of urban investment entities by collecting and processing the public financial report data of listed companies and bond issuers associated with urban investment entities, the public information of enterprises associated with urban investment entities, and the public economic data, regulatory announcements, and news data of governments at all levels in the country and the region where the urban investment entities are located. Based on the data of the urban investment entities themselves, the impact of governments at all levels and public opinion on the credit risk of urban investment entities is fully considered, and the financial status, operating ability, project quality and other aspects of the urban investment entities can be fully reflected. Through the evaluation indicators, investors can have a clearer understanding of the credit status of urban investment entities, so as to make more rational investment decisions and reduce investment risks. By constructing public opinion risk labels, the market public opinion information of urban investment entities can be monitored and analyzed in real time. Through public opinion risk labels, investors can have a more comprehensive understanding of the market image and reputation of urban investment entities, thereby enhancing investment confidence. Based on the correlation analysis results between the first feature and the qualitative indicator, the highly correlated features or indicators are eliminated, and the multicollinearity problem of the multi-factor regression model is avoided, so that the target credit risk assessment model can assess the credit risk more accurately. The first credit risk assessment model and the second credit risk assessment model are combined to construct a target credit risk assessment model, so that the constructed target credit risk assessment model can more accurately assess the credit risk level of the urban investment entity.

[0128] Figure 3 Schematic diagram of a credit risk assessment model generation device provided according to an embodiment of the present invention. Figure 3 As shown, the device comprises:

[0129] The indicator construction module 310 is used to construct evaluation indicators based on data from governments at all levels, urban investment entity data and public opinion data; the evaluation indicators include quantitative indicators and qualitative indicators; the qualitative indicators at least include public opinion risk labels; the public opinion risk labels are obtained by performing natural language processing on the public opinion data; wherein the process of determining the public opinion risk labels includes: performing entity recognition and relationship extraction on the initial news data to obtain target news data associated with the target urban investment entity; performing text classification on the target news data to determine the target risk type of the target news data; based on the association relationship between the pre-set risk type and the risk label, determining the candidate risk labels included in the target risk type; and determining the target risk label of the target news data as the public opinion risk label of the target urban investment entity according to the text similarity between the target news data and the historical news data associated with the candidate risk label.

[0130] The indicator conversion module 320 is used to determine a data conversion model based on the average value and standard deviation of the quantitative indicator in the sample data; and to convert the quantitative indicator into a risk factor based on the data conversion model to obtain a first feature;

[0131] A feature screening module 330, configured to screen a target feature from the first feature and the qualitative indicator according to a correlation analysis result of the first feature and a correlation analysis result of the qualitative indicator;

[0132] The assessment model determination module 340 is used to construct a target credit risk assessment model according to the target characteristics and determine the parameters of the target credit risk assessment model based on sample data.

[0133] The credit risk assessment model generation device provided in the embodiment of the present invention can execute the credit risk assessment model generation method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0134] Optionally, the indicator conversion module is specifically used for:

[0135] For any quantitative indicator, a first reference value is determined based on a preset first confidence level and an average value and a standard deviation of the quantitative indicator; a second reference value is determined based on a preset second confidence level and an average value and a standard deviation of the quantitative indicator;

[0136] The data conversion model is determined based on the first confidence level, the first reference value, the second confidence level, and the second reference value.

[0137] Optionally, the feature screening module includes:

[0138] A first feature screening unit, configured to select a feature from any group of features in the first feature as a candidate feature if the Pearson correlation coefficient between the features in the group is greater than or equal to a correlation threshold;

[0139] The qualitative feature screening unit is used to use the importance of discrete values ​​in the qualitative indicators as the rank of the sample data under the qualitative indicators. If the sample data have the same rank under the qualitative indicators, the correlation between the qualitative indicators is determined according to the Pearson correlation coefficient; if the ranks of the sample data under any two qualitative indicators are different, the correlation between the two qualitative indicators is determined according to the Spearman correlation coefficient; if the correlation coefficient between any group of indicators in the qualitative indicators is greater than or equal to the correlation threshold, a qualitative indicator is selected from the group of indicators as a candidate feature;

[0140] The target feature determination unit is used to take the candidate feature and the feature whose correlation coefficient is less than the correlation threshold as the target feature.

[0141] Optionally, the first feature screening unit is specifically used to:

[0142] The degree of influence of any feature in the set of features on the credit score is determined, and a candidate feature is selected from the set of features according to the degree of influence.

[0143] Optionally, the evaluation model determination module is specifically used for:

[0144] Construct the first credit risk assessment model based on the target characteristics of governments at all levels;

[0145] Construct a second credit risk assessment model based on the target characteristics of the municipal investment entity;

[0146] The target credit risk assessment model is determined according to the first credit risk assessment model and the second credit risk assessment model.

[0147] Optionally, it also includes a risk label and risk type module, which is used to pre-set the association relationship between each risk type and the risk label, and each risk type includes at least one risk label; wherein the risk label is associated with historical news data.

[0148] The credit risk assessment model generation device further described can also execute the credit risk assessment model generation method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0149] Figure 4 A schematic diagram of an electronic device 40 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0150] like Figure 4As shown, the electronic device 40 includes at least one processor 41, and a memory connected to the at least one processor 41, such as a read-only memory (ROM) 42, a random access memory (RAM) 43, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 41 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 42 or the computer program loaded from the storage unit 48 to the random access memory (RAM) 43. In the RAM 43, various programs and data required for the operation of the electronic device 40 can also be stored. The processor 41, the ROM 42, and the RAM 43 are connected to each other through a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0151] A number of components in the electronic device 40 are connected to the I / O interface 45, including: an input unit 46, such as a keyboard, a mouse, etc.; an output unit 47, such as various types of displays, speakers, etc.; a storage unit 48, such as a disk, an optical disk, etc.; and a communication unit 49, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 49 allows the electronic device 40 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0152] The processor 41 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 41 executes the various methods and processes described above, such as a credit risk assessment model generation method.

[0153] In some embodiments, the credit risk assessment model generation method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 48. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 40 via the ROM 42 and / or the communication unit 49. When the computer program is loaded into the RAM 43 and executed by the processor 41, one or more steps of the credit risk assessment model generation method described above may be performed. Alternatively, in other embodiments, the processor 41 may be configured to execute the credit risk assessment model generation method in any other appropriate manner (e.g., by means of firmware).

[0154] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0155] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0156] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), 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 foregoing.

[0157] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0158] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0159] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0160] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0161] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for generating a credit risk assessment model, characterized in that: The method comprises: Constructing evaluation indicators based on data from governments at all levels, data from urban investment entities, and public opinion data; the evaluation indicators include quantitative indicators and qualitative indicators; the qualitative indicators at least include public opinion risk labels; the public opinion risk labels are obtained by performing natural language processing on the public opinion data; Determine a data conversion model based on the average value and standard deviation of the quantitative indicator in the sample data; perform risk factor conversion on the quantitative indicator based on the data conversion model to obtain a first feature; Selecting a target feature from the first feature and the qualitative indicator according to a correlation analysis result of the first feature and a correlation analysis result of the qualitative indicator; Constructing a target credit risk assessment model according to the target characteristics, and determining parameters of the target credit risk assessment model based on sample data; The process of determining the public opinion risk label includes: Perform entity recognition and relationship extraction on the initial news data to obtain the target news data associated with the target municipal investment entity; Performing text classification on the target news data to determine the target risk type of the target news data; Determine the candidate risk tags included in the target risk type based on the association relationship between the preset risk types and the risk tags; According to the text similarity between the target news data and the historical news data associated with the candidate risk label, determining the target risk label of the target news data as the public opinion risk label of the target urban investment entity; The analysis process of the correlation analysis results of the qualitative indicators is as follows: The importance of discrete values ​​in qualitative indicators is taken as the rank of sample data under qualitative indicators. If the sample data have the same rank under qualitative indicators, the correlation between qualitative indicators is determined according to the Pearson correlation coefficient; if the ranks of sample data under any two qualitative indicators are different, the correlation between the two qualitative indicators is determined according to the Spearman correlation coefficient.

2. The method according to claim 1, characterized in that The method further comprises: An association relationship between each risk type and a risk label is preset, and each risk type includes at least one risk label; wherein the risk label is associated with historical news data.

3. The method according to claim 1, characterized in that For any quantitative indicator, determining the data conversion model based on the average value and standard deviation of the quantitative indicator in the sample data includes: Determine a first reference value based on a preset first confidence level and an average value and a standard deviation of the quantitative indicator; determine a second reference value based on a preset second confidence level and an average value and a standard deviation of the quantitative indicator; The data conversion model is determined based on the first confidence level, the first reference value, the second confidence level, and the second reference value.

4. The method according to claim 1, characterized in that The step of selecting a target feature from the first feature and the qualitative indicator according to the correlation analysis result of the first feature and the correlation analysis result of the qualitative indicator includes: If the Pearson correlation coefficient between any group of features in the first feature is greater than or equal to the correlation threshold, then one feature is selected from the group of features as the candidate feature; If the correlation coefficient between any group of indicators in the qualitative indicators is greater than or equal to the correlation threshold, a qualitative indicator is selected from the group of indicators as a candidate feature; The candidate features and features whose correlation coefficients are less than a correlation threshold are used as the target features.

5. The method according to claim 4, characterized in that If the Pearson correlation coefficient between any group of features in the first feature is greater than or equal to the correlation threshold, then selecting a feature from the group of features as a candidate feature includes: The degree of influence of any feature in the set of features on the credit score is determined, and a candidate feature is selected from the set of features according to the degree of influence.

6. The method according to claim 1, characterized in that The step of constructing a target credit risk assessment model according to the target characteristics and determining parameters of the target credit risk assessment model based on sample data includes: Construct the first credit risk assessment model based on the target characteristics of governments at all levels; Construct a second credit risk assessment model based on the target characteristics of the municipal investment entity; The target credit risk assessment model is determined according to the first credit risk assessment model and the second credit risk assessment model.

7. A credit risk assessment model generation device, characterized in that: The device comprises: An indicator construction module is used to construct evaluation indicators based on data from governments at all levels, urban investment entities and public opinion data; the evaluation indicators include quantitative indicators and qualitative indicators; the qualitative indicators at least include public opinion risk labels; the public opinion risk labels are obtained by performing natural language processing on the public opinion data; wherein the process of determining the public opinion risk labels includes: performing entity recognition and relationship extraction on the initial news data to obtain target news data associated with the target urban investment entity; performing text classification on the target news data to determine the target risk type of the target news data; based on the association relationship between the pre-set risk type and the risk label, determining the candidate risk labels included in the target risk type; and determining the target risk label of the target news data as the public opinion risk label of the target urban investment entity based on the text similarity between the target news data and the historical news data associated with the candidate risk label. An indicator conversion module, used to determine a data conversion model based on the average value and standard deviation of the quantitative indicator in the sample data; and to convert the quantitative indicator into a risk factor based on the data conversion model to obtain a first feature; A feature screening module, used for screening target features from the first features and the qualitative indicators according to the correlation analysis results of the first features and the correlation analysis results of the qualitative indicators; An assessment model determination module, used to construct a target credit risk assessment model according to the target characteristics, and determine the parameters of the target credit risk assessment model based on sample data; The analysis process of the correlation analysis results of the qualitative indicators is as follows: The importance of discrete values ​​in qualitative indicators is taken as the rank of sample data under qualitative indicators. If the sample data have the same rank under qualitative indicators, the correlation between qualitative indicators is determined according to the Pearson correlation coefficient; if the ranks of sample data under any two qualitative indicators are different, the correlation between the two qualitative indicators is determined according to the Spearman correlation coefficient.

8. The device according to claim 7, characterized in that It also includes a risk label and risk type module, which is used to pre-set the association relationship between each risk type and the risk label, and each risk type includes at least one risk label; wherein the risk label is associated with historical news data.

9. An electronic device, characterized in that: The electronic device comprises: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the credit risk assessment model generation method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the credit risk assessment model generation method according to any one of claims 1 to 6 when executed.

Images