Credit rating evaluation method and device and computer equipment

Abstract

The invention relates to a credit rating evaluation method and device and computer equipment. The method comprises the steps of performing grouping processing on a to-be-evaluated new citizen according to user portrait data of the to-be-evaluated new citizen, determining an affiliation grouping category of the to-be-evaluated new citizen, performing credit rating evaluation according to the user portrait data and a credit rating evaluation model of the affiliation grouping category, and obtaining a preliminary credit rating quantized value, and determining a credit rating quantized value of the new citizen to be evaluated according to the preliminary credit rating quantized value. Through adoption of the method, credit rating evaluation can be performed on the new citizens to be evaluated according to the credit rating evaluation model corresponding to the affiliation grouping category of the new citizens to be evaluated, so that the accuracy of obtained initial credit rating evaluation of the new citizens can be improved, and on the basis, the credit rating evaluation efficiency is improved. The final credit rating quantized value can be determined according to the initial credit rating evaluation of the new citizens, so that the accuracy of the obtained credit rating evaluation of the new citizens can be improved to a great extent.

Description

Technical Field

[0001] The present application relates to the field of big data technology, and in particular to a credit rating assessment method, device and computer equipment. Background Art

[0002] With the development of new urbanization technologies, new citizens are not only the main force of urban construction and consumption, but also an important force in promoting economic development. Therefore, it is particularly important for financial service institutions to strengthen financial services for new citizen groups.

[0003] In the related art, financial service institutions mainly evaluate the credit rating of any new citizen through a unified credit rating evaluation algorithm, and selectively provide credit financial services to the new citizen based on the evaluated credit rating.

[0004] However, in the process of evaluating the credit rating of new citizens in the related technology, there is a problem of inaccurate evaluation results. Summary of the invention

[0005] Based on this, it is necessary to provide a credit rating assessment method, device and computer equipment to address the above technical issues, which can improve the accuracy of the assessment results in the process of credit rating assessment of new citizens.

[0006] In a first aspect, an embodiment of the present application provides a credit rating assessment method, the method comprising:

[0007] According to the user portrait data of the new citizens to be evaluated, the new citizens to be evaluated are grouped and the group categories to which the new citizens to be evaluated belong are determined;

[0008] Conduct credit rating assessment based on user profile data and the credit rating assessment model of the grouping category to obtain a preliminary credit rating quantitative value;

[0009] Based on the preliminary credit rating quantitative value, determine the credit rating quantitative value of the new citizen to be evaluated.

[0010] The technical solution in the embodiment of the present application is to group the new citizens to be evaluated according to the user portrait data of the new citizens to be evaluated, determine the group category to which the new citizens to be evaluated belong, perform credit rating assessment according to the user portrait data and the credit rating assessment model of the group category to which they belong, obtain a preliminary credit rating quantitative value, and determine the credit rating quantitative value of the new citizens to be evaluated according to the preliminary credit rating quantitative value; the above method can perform credit rating assessment on the new citizens to be evaluated according to the credit rating assessment model corresponding to the group category to which the new citizens to be evaluated belong, thereby improving the accuracy of the preliminary credit rating assessment of the new citizens obtained, and on this basis, can determine the most accurate credit rating based on the preliminary credit rating assessment of the new citizens. The final quantitative value of the credit rating can greatly improve the accuracy of the credit rating assessment of the new citizens. At the same time, the above method does not require the participation of complex algorithms, which can improve the complexity of the new citizen credit rating assessment method and reduce the amount of data processed in the new citizen credit rating assessment process, thereby improving the speed and efficiency of the new citizen credit rating assessment. Furthermore, the above method uses user portrait data of the new citizens to be assessed to eliminate the data island problem and supplement different types of missing data, making the source data of the credit rating assessment more comprehensive and valuable, that is, making the basic data relied on by the new citizen credit rating assessment more accurate, thereby achieving accurate assessment of the new citizen credit rating.

[0011] In one embodiment, the new citizens to be evaluated are grouped according to the user portrait data of the new citizens to be evaluated, and the group category to which the new citizens to be evaluated belong is determined, including:

[0012] Perform standardization processing on user portrait data to obtain standardized data;

[0013] The standardized data is input into the pre-trained clustering model to obtain the clustering category to which the new citizens to be evaluated belong.

[0014] The technical solution in the embodiment of the present application performs standardized processing on the user portrait data to obtain standardized data, and inputs the standardized data into a pre-trained clustering model to obtain the grouping category to which the new citizen to be evaluated belongs; the above method can first group the new citizens to be evaluated, determine the grouping category to which the new citizens to be evaluated belong, so as to provide reference information for further accurate credit rating assessment of the new citizens to be evaluated.

[0015] In one embodiment, the user portrait data includes numerical variable data and categorical variable data; performing standardization processing on the user portrait data to obtain standardized data includes:

[0016] Normalizing the numerical variable data to obtain normalized data; and encoding the categorical variable data to obtain encoded data;

[0017] The normalized data and the encoded data are determined as the normalized data.

[0018] The technical solution in the embodiment of the present application performs normalization processing on numerical variable data to obtain normalized data, and performs encoding processing on categorical variable data to obtain encoded data, and determines the normalized data and the encoded data as standardized data; the above method can perform standardization processing on the user portrait data of the new citizens to be evaluated, so as to prepare for further accurate grouping of the new citizens to be evaluated, and the standardization processing does not require the participation of complex algorithms, thereby reducing the amount of data involved in the standardization processing and improving the speed and efficiency of the standardization processing.

[0019] In one embodiment, before grouping the new citizens to be evaluated according to the user portrait data of the new citizens to be evaluated, the method further includes:

[0020] Obtain training data of multiple sample new citizens;

[0021] Perform principal component analysis on the training data to obtain the analysis results;

[0022] According to the analysis result, the first valid data is selected from the training data;

[0023] The initial clustering model is trained according to the first valid data to generate a clustering model.

[0024] The technical solution in the embodiment of the present application obtains training data of multiple sample new citizens, performs principal component analysis on the training data to obtain analysis results, screens first valid data from the training data according to the analysis results, and trains an initial clustering model based on the first valid data to generate a clustering model; the above method can train an optimal clustering model to prepare for the subsequent rapid clustering of new citizens to be evaluated based on the pre-trained clustering model, and the training process does not require additional operations, and the training process is relatively simple; at the same time, the above method can screen out the first valid data corresponding to the valid variables from the training data, and train the initial clustering model based on the first valid data, so as to obtain a clustering model with higher accuracy.

[0025] In one embodiment, the process of generating the credit rating assessment model includes:

[0026] Perform variable selection on the training data of multiple sample new citizens to obtain the second effective data;

[0027] The initial evaluation model is trained according to the second valid data to obtain a credit rating evaluation model.

[0028] The technical solution in the embodiment of the present application performs variable selection on the training data of multiple sample new citizens to obtain second valid data, and trains the initial evaluation model based on the second valid data to obtain a credit rating evaluation model; the above method can filter out the second valid data corresponding to the valid variables from the training data, and train the initial evaluation model based on the second valid data, so as to obtain a clustering model with higher accuracy; the above method trains a credit rating evaluation model based on a logistic regression algorithm, and realizes credit rating evaluation in combination with a clustering model, so as to improve the interpretability of the evaluated credit rating.

[0029] In one embodiment, variable selection is performed on the training data of multiple sample new citizens to obtain second valid data, including:

[0030] According to the training data, each variable in the training data is coded with evidence weight to obtain the converted data of each variable;

[0031] Determine the quantitative value of information value corresponding to each variable based on the converted data of each variable;

[0032] Determine at least one valid variable among the variables according to the quantitative value of the information value of each variable;

[0033] The data of each valid variable is obtained from the training data, and the data of each valid variable is determined as the second valid data.

[0034] The technical solution in the embodiment of the present application performs evidence weight encoding processing on each variable in the training data according to the training data to obtain converted data of each variable, determines the information value quantification value corresponding to each variable according to the converted data of each variable, determines at least one valid variable in each variable according to the information value quantification value of each variable, obtains the data of each valid variable from the training data, and determines the data of each valid variable as the second valid data; the above method can filter out the data corresponding to the valid variables from the training data to train the credit rating assessment model, so as to prepare for further training of a credit rating assessment model with higher accuracy, and the processing process is relatively simple and does not require the participation of complex algorithms, thereby improving the speed and efficiency of obtaining the second valid data.

[0035] In one embodiment, determining the information value quantification value corresponding to each variable according to the converted data of each variable includes:

[0036] For any variable, the transformed data of the variable are grouped to obtain multiple data sets;

[0037] For any set of data sets, the data sets are converted into evidence weights to obtain the corresponding evidence weight values;

[0038] The variable value is evaluated according to the evidence weight value of each data set to determine the quantitative value of the information value corresponding to the variable.

[0039] The technical solution in the embodiment of the present application is to group the converted data of any variable to obtain multiple data sets, and for any data set, perform evidence weight conversion on the data set to obtain a corresponding evidence weight value, and evaluate the value of the variable according to the evidence weight value of each data set to determine the quantitative value of the information value corresponding to the variable; the above method can calculate the corresponding quantitative value of the information value for any variable in the training data, so as to prepare for further screening of effective variables, and the processing process is relatively simple and does not require the participation of complex algorithms, thereby improving the speed and efficiency of the process.

[0040] In one embodiment, determining the target credit rating quantitative value of the new citizen to be evaluated based on the preliminary credit rating quantitative value includes:

[0041] The preliminary credit rating quantified value is input into the comprehensive evaluation model of the credit rating, and the target credit rating quantified value is output; the comprehensive evaluation model of the credit rating is a model trained based on the sample quantified values ​​of the credit ratings of new users in different grouping categories.

[0042] The technical solution in the embodiment of the present application can further determine the target credit rating quantitative value of the new citizen to be evaluated on the basis of obtaining the preliminary credit rating quantitative value of the new citizen to be evaluated, so as to complete the credit rating evaluation of the new citizen to be evaluated, so that the credit rating evaluation results of the new citizen are more accurate, and on this basis, financial institutions can provide credit financial services to new citizens more reasonably and accurately, thereby reducing the occurrence rate of erroneous provision or omission.

[0043] In a second aspect, an embodiment of the present application provides a credit rating assessment device, the device comprising:

[0044] A grouping processing module is used to group the new citizens to be evaluated according to the user portrait data of the new citizens to be evaluated, and determine the grouping category to which the new citizens to be evaluated belong;

[0045] An evaluation module is used to evaluate the credit rating based on the user portrait data and the credit rating evaluation model of the grouping category to obtain a preliminary credit rating quantitative value;

[0046] The determination module is used to determine the target credit rating quantitative value of the new citizen to be evaluated based on the preliminary credit rating quantitative value.

[0047] In a third aspect, an embodiment of the present application further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method of any one of the embodiments of the first aspect when executing the computer program.

[0048] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method of any embodiment in the above-mentioned first aspect are implemented.

[0049] In a fifth aspect, an embodiment of the present application further provides a computer program product, the computer program product comprising a computer program, which, when executed by a processor, implements the steps of the method of any embodiment of the first aspect above.

[0050] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 An application environment diagram of a credit rating assessment method in an embodiment;

[0052] Figure 2 A flow chart of a credit rating assessment method in one embodiment;

[0053] Figure 3 A flow chart of a credit rating evaluation method in another embodiment;

[0054] Figure 4 A flow chart of a credit rating evaluation method in another embodiment;

[0055] Figure 5 A flow chart of a credit rating evaluation method in another embodiment;

[0056] Figure 6 A flow chart of a credit rating evaluation method in another embodiment;

[0057] Figure 7 A flow chart of a credit rating evaluation method in another embodiment;

[0058] Figure 8 A flow chart of a credit rating evaluation method in another embodiment;

[0059] Fig. 9 is a structural block diagram of a credit rating evaluation device in one embodiment;

[0060] Fig.10 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0062] In the financial field, it is particularly important for financial service institutions to strengthen financial services for the new citizen group. In the related art, financial service institutions mainly evaluate the credit rating of any new citizen through a unified credit rating evaluation algorithm, and selectively provide credit financial services to the new citizen based on the evaluated credit rating. However, in the process of evaluating the credit rating of new citizens in the related art, there is a problem of inaccurate evaluation results. Based on this, the embodiment of the present application provides a credit rating evaluation method, which can improve the accuracy of the evaluation results in the process of evaluating the credit rating of new citizens.

[0063] The credit rating evaluation method provided in the embodiment of the present application can be applied to Figure 1 The application environment diagram shown. The application environment includes a computer device and multiple data management devices. Different data management devices are data management devices for different departments, which are used to store various types of data for different users. Optionally, there can be a communication connection between the computer device and each database system, and the communication method can be Wi-Fi, mobile network or Bluetooth connection, etc. Among them, the computer device can be a personal computer, laptop computer, desktop computer, smart phone and other electronic device with data processing function; each data management device can be a computer device, storage device, database system and other device with data management function; this embodiment does not limit the specific form of the computer device and each data management device. Among them, Figure 1 The example is illustrated by taking three data management devices (data management device 1, data management device 2 and data management device 3). The specific process of the credit rating evaluation method is introduced below by taking a computer device as the execution subject of the credit rating evaluation method.

[0064] like Figure 2 FIG. 1 is a flow chart of a credit rating evaluation method provided in an embodiment of the present application, and the method can be implemented by the following steps:

[0065] S100. Grouping the new citizens to be evaluated based on their user portrait data to determine the grouping category to which they belong.

[0066] In practical applications, citizens with characteristics such as unstable residence, poor job stability, single lifestyle, and low participation in social security can be called new citizens. In the embodiment of the present application, the user portrait data of the new citizen to be evaluated may include data from different sources of the new citizen to be evaluated, including public security data, government data, social data, industry data, Internet data, etc.

[0067] Among them, the above-mentioned public security data may include the basic information, residence information, social information, criminal record information, traffic violation information, etc. of the new citizens to be evaluated; government data may include social security payment information, provident fund payment information, education information, industrial and commercial information, tax information, judicial information, etc. of the new citizens to be evaluated; social data may include administrative law enforcement information, telecommunications payment information, professional skills information, social service activity information, etc. of the new citizens to be evaluated; industry data may include bank transaction information, deposit data, historical credit information, etc. of the new citizens to be evaluated; Internet data may include Internet financial consumption information, Internet financial historical credit information, etc. of the new citizens to be evaluated.

[0068] Specifically, the computer device can obtain various types of data of the new citizens to be evaluated, that is, user portrait data, from various data management devices, and group the new citizens to be evaluated to obtain the grouping categories to which the new citizens to be evaluated belong.

[0069] In one embodiment, the method of grouping the new citizens to be evaluated may be to perform comparison processing and / or matching processing on the user portrait data of the new citizens to be evaluated to obtain the group category to which the new citizens to be evaluated belong.

[0070] In another embodiment, the method of grouping the new citizens to be evaluated may be to pre-train an algorithm model, and then input the user portrait data of the new citizens to be evaluated into the algorithm model, and the algorithm model outputs the group category to which the new citizens to be evaluated belong.

[0071] It should be noted here that the above-mentioned new citizens to be evaluated may belong to groups such as graduated white-collar workers, entrepreneurial youth, and migrant workers.

[0072] S200: Perform credit rating assessment based on user portrait data and a credit rating assessment model belonging to a grouping category to obtain a preliminary credit rating quantitative value.

[0073] In practical applications, a credit rating assessment model corresponding to the grouping category of the new citizen to be assessed can be obtained, and the user portrait data of the new citizen to be assessed can be input into the credit rating assessment model corresponding to the grouping category, and the credit rating assessment model outputs a preliminary credit rating quantitative value.

[0074] S300: Determine the credit rating quantification value of the new citizen to be evaluated based on the preliminary credit rating quantification value.

[0075] The computer device may perform arithmetic operations, conversion processing, matching processing, etc. on the preliminary credit rating quantization value to obtain the credit rating quantization value of the new citizen to be evaluated. Optionally, the arithmetic operations may include at least one of addition, subtraction, multiplication, division, logarithm, exponential, etc.

[0076] The technical solution in the embodiment of the present application is to group the new citizens to be evaluated according to the user portrait data of the new citizens to be evaluated, determine the group category to which the new citizens to be evaluated belong, perform credit rating assessment according to the user portrait data and the credit rating assessment model of the group category to which they belong, obtain a preliminary credit rating quantitative value, and determine the credit rating quantitative value of the new citizens to be evaluated according to the preliminary credit rating quantitative value; the above method can perform credit rating assessment on the new citizens to be evaluated according to the credit rating assessment model corresponding to the group category to which the new citizens to be evaluated belong, thereby improving the accuracy of the preliminary credit rating assessment of the new citizens obtained, and on this basis, can determine the most accurate credit rating based on the preliminary credit rating assessment of the new citizens. The final quantitative value of the credit rating can greatly improve the accuracy of the credit rating assessment of the new citizens. At the same time, the above method does not require the participation of complex algorithms, which can improve the complexity of the new citizen credit rating assessment method and reduce the amount of data processed in the new citizen credit rating assessment process, thereby improving the speed and efficiency of the new citizen credit rating assessment. Furthermore, the above method uses user portrait data of the new citizens to be assessed to eliminate the data island problem and supplement different types of missing data, making the source data of the credit rating assessment more comprehensive and valuable, that is, making the basic data relied on by the new citizen credit rating assessment more accurate, thereby achieving accurate assessment of the new citizen credit rating.

[0077] In one embodiment, Figure 3 As shown, the step of grouping the new citizens to be evaluated according to the user portrait data of the new citizens to be evaluated and determining the group category to which the new citizens to be evaluated belong in the above S200 can be implemented in the following manner:

[0078] S210, performing standardization processing on the user portrait data to obtain standardized data.

[0079] In practical applications, the computer device can pre-train a standardized model, and then input the user portrait data of the new user to be evaluated into the standardized model, and the standardized model outputs the standardized data. Optionally, the standardized model can be at least one of a convolutional neural network model, a fully connected neural network model, a long short-term memory neural network model, a recurrent neural network model, a residual neural network model, etc.

[0080] In the embodiment of the present application, the computer device can perform data exploration on the user portrait data to determine the data type, data distribution and variable type to which the data belongs, find outliers or quality problems in the user portrait data, and achieve full understanding of the user portrait data. Further, the user portrait data can be standardized based on the data exploration results.

[0081] The data exploration method may include feature matching, keyword extraction, regular expression and other processing methods. For example, data exploration may identify incorrect ID card numbers, mobile phone numbers and the like based on regular expressions and other methods.

[0082] Optionally, the processing specifically included in the above-mentioned standardization processing can be determined based on abnormal problems determined during the data exploration process, and the abnormal problems may include inconsistent data standards, inconsistent data specifications, inconsistent data formats, dirty data, data non-compliance, data errors, data duplication, data omissions, and no correlation between data in the user portrait data.

[0083] If the abnormal problem includes inconsistent data standards and / or inconsistent data specifications in the user portrait data, the standardization process may include data definition normalization processing; data definition normalization processing is to normalize the naming rules, data types, lengths, etc. of the data in the user portrait data to determine the consistency of the data. If the abnormal problem includes inconsistent data formats in the user portrait data, the standardization process may include data format conversion processing; data format conversion processing is to convert various types of data in the user portrait data into a unified format according to the data standard content. Various types of data may include date and time standardization, data types, unit unification, non-standard codes, etc.; for example, in accordance with the provisions of the "Gender Code of People", male or male in the user portrait data is uniformly converted into code "1", and female or female is uniformly converted into code "2". If the abnormal problem includes dirty data or non-compliant data in the user portrait data, the standardization process may include data unification, which is to use data processing strategies to unify dirty data and non-compliant data in the user portrait data. For example, through the unified mobile phone number strategy, data such as "+86 139-1234-XXXX" can be processed into a unified "1391234XXXX". If the abnormal problem includes data errors, data inconsistencies, data duplications and / or data omissions in the user portrait data, the standardization process may include data cleaning; data cleaning refers to the process of supplementing missing values ​​in the user portrait data, correcting erroneous data fields, and removing duplicate data records. For example, if the basic information of the new citizen to be evaluated lacks gender and date of birth data during the data collection process, the gender and date of birth can be extracted and supplemented through the ID card number of the person. If the abnormal problem includes the lack of correlation between the data in the user portrait data, the standardization processing may include multi-source data fusion or association. Multi-source data fusion or association refers to associating and integrating multiple source data in the user portrait data through certain rules and supplementing each other to improve the quality, accuracy and consistency of the data. For example, by using the identity number of the new citizen to be evaluated as the primary key, the basic information of the new citizen to be evaluated can be associated with the provident fund payment information, and the income level information of the new citizen to be evaluated can be inferred through the provident fund payment base.

[0084] S220, input the standardized data into a pre-trained clustering model to obtain the clustering category to which the new citizen to be evaluated belongs.

[0085] Specifically, the computer device can input the standardized data into a pre-trained clustering model, and the clustering model divides the standardized data into the belonging clustering categories to determine the belonging clustering category of the new citizen to be evaluated. The above-mentioned clustering model can be an unsupervised clustering model, and the unsupervised clustering model can be a K-means clustering model, a Gaussian mixture model, a density-based spatial clustering model, etc.; in the embodiment of the present application, the above-mentioned clustering model can be a hierarchical clustering model, and the hierarchical clustering model can use the sum of squared deviations method and the Euclidean distance method to participate in the processing.

[0086] The technical solution in the embodiment of the present application performs standardized processing on the user portrait data to obtain standardized data, and inputs the standardized data into a pre-trained clustering model to obtain the grouping category to which the new citizen to be evaluated belongs; the above method can first group the new citizens to be evaluated, determine the grouping category to which the new citizens to be evaluated belong, so as to provide reference information for further accurate credit rating assessment of the new citizens to be evaluated.

[0087] The following is an explanation of the process of obtaining standardized data by performing standardization processing on the user portrait data. In one embodiment, the user portrait data includes numerical variable data and categorical variable data; Figure 4 As shown, the steps in the above S210 may include:

[0088] S211, normalizing the numerical variable data to obtain normalized data; and encoding the categorical variable data to obtain encoded data.

[0089] Specifically, the computer device can classify the user portrait data of the new user to be evaluated to obtain numerical variable data and categorical variable data.

[0090] In practical applications, computer equipment can perform normalization processing on numerical variable data to obtain normalized data, and can also perform encoding processing on categorical variable data to obtain encoded data.

[0091] S212: Determine the normalized data and the encoded data as standardized data.

[0092] In the embodiment of the present application, the normalized data and the encoded data may be determined as the standardized data.

[0093] The technical solution in the embodiment of the present application performs normalization processing on numerical variable data to obtain normalized data, and performs encoding processing on categorical variable data to obtain encoded data, and determines the normalized data and the encoded data as standardized data; the above method can perform standardization processing on the user portrait data of the new citizens to be evaluated, so as to prepare for further accurate grouping of the new citizens to be evaluated, and the standardization processing does not require the participation of complex algorithms, thereby reducing the amount of data involved in the standardization processing and improving the speed and efficiency of the standardization processing.

[0094] In practical applications, before executing the steps in S100 above, it is necessary to construct a clustering model. The process of generating a clustering model is described below. Figure 5 As shown, the above method may further include the following steps:

[0095] S400: Obtain training data of multiple sample new citizens.

[0096] In practical applications, the computer device can obtain different original source data of multiple sample citizens in the target area from each data management device, and then filter out different source data of multiple sample citizens (i.e., effective source data or user portrait data) from the different original source data of multiple sample citizens, and then filter out user portrait data of multiple sample new citizens from the user portrait data of multiple sample citizens, and divide the user portrait data of multiple sample new citizens into training sets and test sets, and determine training data based on the training sets. Optionally, the multiple sample citizens may include new citizens and permanent residents in the target area.

[0097] It should be noted here that if the proportion of good users and bad users in the label data of the user portrait data of multiple sample new citizens is unbalanced, when dividing the training set and the test set from the user portrait data of multiple sample new citizens, the user portrait data of multiple sample new citizens should be balanced first to obtain a balanced training set, i.e., training data. In this way, the clustering model can be trained by the training data to improve the robustness of the trained clustering model.

[0098] In practical applications, the data in the divided training set and test set can also be standardized to obtain a standardized training set and a standardized test set, and the data in the standardized training set can be determined as training data, so that the deviation and span between different variables in the training data are small, so as to improve the accuracy of the training results. For numerical variables, the variable standardization can be Z-score standardization; for example, the numerical variables ["period", "credit_limit", "address", "age", "credits_num", "income_installment_rate", "dependents"] are subjected to Z-score standardization to scale the data to [0, 1]. At the same time, for categorical variables, the variable standardization can be normalization processing; for example, when the categorical variable is working years [person_employee], the working years can be encoded and the results are as follows: working years of 0 means unemployed, working years of 1 means less than 1 year, working years of 2 means 1 to 4 years, working years of 3 means 4 to 7 years, and working years of 4 means more than 7 years; for categorical variables without size meaning, one-hot encoding can be performed. For example, the result of hot encoding of gender code is as follows: 0 means male, 1 means female.

[0099] Among them, the different original source data of each sample new citizen may include multiple types of information of the sample new citizen, namely basic information, public information and credit information, as shown in Table 1; the user portrait data of each sample new citizen may include the basic information and credit information of the sample new citizen. Among them, the basic information, public information and credit information may all include corresponding secondary information, and each secondary information may include corresponding data items.

[0100] Table 1

[0101]

[0102]

[0103] S500: Perform principal component analysis on the training data to obtain analysis results.

[0104] Specifically, the computer device can use principal component analysis to perform principal component analysis on the training data to obtain analysis results. Optionally, the principal component analysis method can be a principal component analysis method based on a covariance matrix, a principal component analysis method based on a correlation matrix, a robust principal component analysis method, etc.

[0105] S600: Filter first valid data from the training data according to the analysis result.

[0106] It should be noted that the above analysis results may include multiple variables in the training data. Specifically, the computer device may filter out multiple main variables, i.e., main components, from the analysis results, and then filter out the training data corresponding to each main variable from the training data as the first valid data. It should be noted that the above main variables may be valid variables among all variables in the training data.

[0107] In practical applications, the user portrait data of multiple sample new citizens can be classified according to different variables to obtain classification results, including 13 categorical variables and 7 numerical variables, as shown in Table 2.

[0108] Table 2

[0109]

[0110]

[0111]

[0112]

[0113] In addition, the first valid data can also be selected from the training data by performing variable selection on the training data. The variable selection process is similar to the step in step S800, and will not be described in detail in this embodiment of the application.

[0114] S700: Train the initial clustering model according to the first valid data to generate a clustering model.

[0115] Furthermore, the initial clustering model can be trained by the first valid data to generate a clustering model. The first valid data can be input into the initial clustering model to output the predicted clustering result, and then the loss value between the predicted clustering result and the standard clustering result is obtained. If the loss value does not meet the standard, the model parameters of the initial clustering model are updated, and then the first valid data is input into the updated initial clustering model for cyclic iterative processing. If the loss value meets the standard or the number of iterations reaches a preset number threshold, the initial clustering model at this time is determined as the clustering model.

[0116] After the first valid data is input into the initial clustering model, the initial clustering model can output data sets corresponding to attribute sets belonging to multiple clustering categories. Taking 3 categories as an example, the following Table 3 shows the attribute sets corresponding to different clustering categories.

[0117] Table 3

[0118]

[0119] The technical solution in the embodiment of the present application obtains training data of multiple sample new citizens, performs principal component analysis on the training data to obtain analysis results, screens first valid data from the training data according to the analysis results, and trains an initial clustering model based on the first valid data to generate a clustering model; the above method can train an optimal clustering model to prepare for the subsequent rapid clustering of new citizens to be evaluated based on the pre-trained clustering model, and the training process does not require additional operations, and the training process is relatively simple; at the same time, the above method can screen out the first valid data corresponding to the valid variables from the training data, and train the initial clustering model based on the first valid data, so as to obtain a clustering model with higher accuracy.

[0120] In one embodiment, Figure 6 As shown, the generation process of the above credit rating assessment model may include:

[0121] S800: Perform variable selection on the training data of multiple sample new citizens to obtain second valid data.

[0122] In practical applications, the computer device can perform feature extraction on the training data to perform variable selection on the training data to obtain second valid data of multiple valid variables in the training data.

[0123] In addition, the computer device can pre-train an algorithm model, and then input the training data of multiple sample new citizens into the algorithm model, and the algorithm model outputs the second valid data.

[0124] Among them, the method of performing variable selection on the training data of multiple sample new citizens to obtain the second valid data can also be similar to the steps in steps S500-S600, which will not be elaborated in this embodiment of the application.

[0125] S900: Train the initial evaluation model according to the second valid data to obtain a credit rating evaluation model.

[0126] Furthermore, the second valid data can be input into the initial evaluation model to train the initial evaluation model and obtain the optimal credit rating evaluation model. Optionally, the initial evaluation model can be at least one of a convolutional neural network model, a fully connected neural network model, a long short-term memory neural network model, a residual neural network model, a recurrent neural network model, a decision tree, etc. It should be noted that the credit rating evaluation model is used to predict whether a new citizen will have a credit rating corresponding to overdue behavior.

[0127] In the embodiment of the present application, the above initial evaluation model is constructed based on the logistic regression algorithm. The construction process of the initial evaluation model is described below: a linear regression function z=θ0+θ1x1+θ2x2+...+θ m x m =θ T X(X includes x1,...,x m represents the valid data corresponding to each valid variable, that is, the second valid data), and then substitute the second valid data into the linear regression function to obtain z, and then substitute z into the sigmoid function to obtain the initial evaluation model The probability of the initial evaluation model output P1 is equal to in, P1 is greater than or equal to 0.5, θ T If X is greater than or equal to 0, it is determined that the sample new citizen has no overdue credit, that is, the predicted credit rating of the sample new citizen is 1, that is, y(X) is equal to 1; P1 is less than 0.5, θ T If X is less than 0, it is determined that the sample new citizen has been overdue, that is, the predicted credit rating of the sample new citizen is 0, that is, y(X) is equal to 0.

[0128] The technical solution in the embodiment of the present application performs variable selection on the training data of multiple sample new citizens to obtain second valid data, and trains the initial evaluation model based on the second valid data to obtain a credit rating evaluation model; the above method can filter out the second valid data corresponding to the valid variables from the training data, and train the initial evaluation model based on the second valid data, so as to obtain a clustering model with higher accuracy; the above method trains a credit rating evaluation model based on a logistic regression algorithm, and realizes credit rating evaluation in combination with a clustering model, so as to improve the interpretability of the evaluated credit rating.

[0129] The following is an explanation of the process of obtaining the second valid data by performing variable selection on the training data of multiple sample new citizens. Figure 7 As shown, the steps in the above S800 may include:

[0130] S810. Perform evidence weight encoding processing on each variable in the training data according to the training data to obtain converted data of each variable.

[0131] In practical applications, for any variable in the training data, the computer device can perform evidence weight encoding processing on each variable in the training data according to the training data to obtain converted data of each variable.

[0132] In one embodiment, an algorithm model may be pre-trained, and then the data corresponding to each variable in the training data may be encoded using evidence weights to obtain converted data for each variable.

[0133] S820. Determine the information value quantization value corresponding to each variable according to the converted data of each variable.

[0134] Specifically, for any variable, the computer device can determine the information value quantization value corresponding to the variable for the converted data of the variable. In the embodiment of the present application, the above information value quantization value can be expressed as an IV value.

[0135] S830. Determine at least one valid variable among the variables according to the quantified value of the information value of each variable.

[0136] Furthermore, the information value quantification value of each variable can be compared with a preset threshold range, and the corresponding variable whose information value quantification value is within the preset threshold range can be screened out according to the comparison result and determined as at least one valid variable.

[0137] Optionally, the preset threshold range may be an interval range, which may be user-defined or determined based on historical experience, and this embodiment of the present application does not limit this. In the embodiment of the present application, the preset threshold range may be an interval range of 0.3 to 0.5.

[0138] S840. Acquire data of each valid variable from the training data, and determine the data of each valid variable as second valid data.

[0139] In an embodiment of the present application, the computer device may obtain data of each valid variable from the training data and determine it as the second valid data.

[0140] The technical solution in the embodiment of the present application performs evidence weight encoding processing on each variable in the training data according to the training data to obtain converted data of each variable, determines the information value quantification value corresponding to each variable according to the converted data of each variable, determines at least one valid variable in each variable according to the information value quantification value of each variable, obtains the data of each valid variable from the training data, and determines the data of each valid variable as the second valid data; the above method can filter out the data corresponding to the valid variables from the training data to train the credit rating assessment model, so as to prepare for further training of a credit rating assessment model with higher accuracy, and the processing process is relatively simple and does not require the participation of complex algorithms, thereby improving the speed and efficiency of obtaining the second valid data.

[0141] In one embodiment, if Figure 8As shown, the step of determining the information value quantization value corresponding to each variable according to the converted data of each variable in the above S820 can be implemented in the following way:

[0142] S821. For any variable, group the converted data of the variable to obtain multiple data sets.

[0143] For any variable, the computer device can group the converted data of the variable to divide the converted data of the variable into any number of groups to obtain multiple data sets. Optionally, the grouping here can be understood as a process of discretizing continuous variables.

[0144] In the embodiment of the present application, for any variable, the computer device can group the converted data of the variable according to the preset grouping number corresponding to the variable to obtain multiple data sets corresponding to the variable. For any variable, if the variable is a numerical variable, such as age, the age can be divided into multiple groups such as [10, 20], [20, 30], [30 greater than 30], etc.

[0145] S822. For any set of data sets, perform evidence weight conversion on the data sets to obtain corresponding evidence weight values.

[0146] In practical applications, for any set of data sets corresponding to any variable, a computer device can perform evidence weight conversion processing on the data set to obtain the evidence weight value corresponding to the data set.

[0147] Specifically, for any set of data sets i corresponding to any variable, the evidence weight conversion process can be described as follows: first obtain the number y of good users in the data set i i and the number of bad users n in the dataset i , and get the number of good users y in all group data sets corresponding to this variable T The number of bad users n in all group data sets corresponding to this variable T , and then get the number y of good users in the data set i i The number of good users in all group data sets y T The proportion p yi (ie y i / y T ) and the number of bad users n in the dataset i i The number of bad users in all group data sets n T The proportion p ni (i.e. n i / n T ), further, for p yi and p ni Take the logarithm to get the weight of evidence value WOE corresponding to the group of data sets i i .

[0148] The above WOE i The expression can be:

[0149]

[0150] Among them, WOE i It can also represent the difference between the ratio of the number of good users to the number of bad users in the data set i of any variable and the ratio of the number of good users to the number of bad users in all data sets of the variable. i The larger the absolute value of , the greater the difference, and the greater the influence of the training data of good users in the data set i on the credit rating assessment; WOE i The smaller the absolute value of , the smaller the difference, and the smaller the impact of the training data of good users in the data set i on the credit rating assessment; WOE i Equal to 0, indicating that the data set i is not capable of assessing credit rating.

[0151] S823. Evaluate the value of the variable according to the evidence weight value of each data set to determine the quantitative value of the information value corresponding to the variable.

[0152] Specifically, for any set of data sets i, the WOE of the evidence weight of the set of data sets i can be calculated as i Perform variable value evaluation to obtain the information value quantification value IV of the group of data sets i i Furthermore, for any variable, the information value of each set of data sets i corresponding to the variable can be quantified as IV i Sum them together to get the quantitative value IV of the information value corresponding to the variable.

[0153] IV above i The expressions for and IV are as follows:

[0154]

[0155] Among them, m represents the number of groups of data sets corresponding to any variable.

[0156] The technical solution in the embodiment of the present application is to group the converted data of any variable to obtain multiple data sets, and for any data set, perform evidence weight conversion on the data set to obtain a corresponding evidence weight value, and evaluate the value of the variable according to the evidence weight value of each data set to determine the quantitative value of the information value corresponding to the variable; the above method can calculate the corresponding quantitative value of the information value for any variable in the training data, so as to prepare for further screening of effective variables, and the processing process is relatively simple and does not require the participation of complex algorithms, thereby improving the speed and efficiency of the process.

[0157] In one embodiment, the step of determining the target credit rating quantitative value of the new citizen to be evaluated based on the preliminary credit rating quantitative value in S300 may include: inputting the preliminary credit rating quantitative value into a comprehensive evaluation model of credit rating, and outputting the target credit rating quantitative value. The comprehensive evaluation model of credit rating is a model trained based on sample quantitative values ​​of credit ratings of new users of different grouping categories.

[0158] In practical applications, the above comprehensive evaluation model can be obtained by training the initial comprehensive evaluation model based on the sample quantized values ​​of the credit ratings of new users in different grouping categories. Optionally, the above initial comprehensive evaluation model can be implemented by at least one of a convolutional neural network model, a fully connected neural network model, a long short-term memory neural network model, a recurrent neural network model, a residual neural network model, etc.

[0159] Furthermore, the quantitative value of the preliminary credit rating of the new citizen to be evaluated can be input into the comprehensive evaluation model of the credit rating, and the quantitative value of the target credit rating of the new citizen to be evaluated can be output.

[0160] The technical solution in the embodiment of the present application can further determine the target credit rating quantitative value of the new citizen to be evaluated on the basis of obtaining the preliminary credit rating quantitative value of the new citizen to be evaluated, so as to complete the credit rating evaluation of the new citizen to be evaluated, so that the credit rating evaluation results of the new citizen are more accurate, and on this basis, financial institutions can provide credit financial services to new citizens more reasonably and accurately, thereby reducing the occurrence rate of erroneous provision or omission.

[0161] In one embodiment, the present application also provides a credit rating assessment method, which is applied to a computer device, and the method includes the following process:

[0162] (1) normalizing the numerical variable data to obtain normalized data; and encoding the categorical variable data to obtain encoded data;

[0163] (2) determining the normalized data and the coded data as the standardized data;

[0164] (3) Input the standardized data into the pre-trained clustering model to obtain the clustering category to which the new citizens to be evaluated belong;

[0165] (4) Conducting credit rating assessment based on user profile data and the credit rating assessment model of the grouping category to obtain a preliminary credit rating quantitative value;

[0166] (5) Inputting the preliminary credit rating quantified value into the comprehensive evaluation model of the credit rating, and outputting the target credit rating quantified value; the comprehensive evaluation model of the credit rating is a model trained based on the sample quantified values ​​of the credit ratings of new users of different grouping categories;

[0167] The generation process of the clustering model in the above step (3) includes:

[0168] (31) Obtaining training data of multiple sample new citizens;

[0169] (32) Performing principal component analysis on the training data to obtain analysis results;

[0170] (33) selecting first valid data from the training data according to the analysis result;

[0171] (34) training the initial clustering model according to the first valid data to generate a clustering model;

[0172] The generation process of the credit rating assessment model for the grouping category in the above step (4) includes:

[0173] (41) Based on the training data of multiple sample new citizens, each variable in the training data is encoded with evidence weight to obtain the converted data of each variable;

[0174] (42) For any variable, the transformed data of the variable are grouped to obtain multiple data sets;

[0175] (43) For any set of data sets, the data sets are converted into evidence weights to obtain corresponding evidence weight values;

[0176] (44) Evaluate the variable value based on the evidence weight of each data set and determine the quantitative value of the information value corresponding to the variable;

[0177] (45) determining at least one valid variable among the variables according to the quantified value of the information value of each variable;

[0178] (46) obtaining data of each valid variable from the training data, and determining the data of each valid variable as second valid data;

[0179] (47) Based on the second valid data, the initial evaluation model is trained to obtain a credit rating evaluation model.

[0180] The execution process of the above (1) to (5) can be specifically referred to the description of the above embodiment. The implementation principle and technical effect are similar and will not be repeated here.

[0181] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0182] Based on the same inventive concept, the embodiment of the present application also provides a credit rating assessment device for implementing the credit rating assessment method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more credit rating assessment device embodiments provided below can refer to the limitations of the credit rating assessment method above, and will not be repeated here.

[0183] In one embodiment, Fig. 9 This is a schematic diagram of the structure of a credit rating evaluation device in one embodiment of the present application. The credit rating evaluation device provided in the embodiment of the present application can be applied to a computer device. Fig. 9 As shown, the credit rating evaluation device of the embodiment of the present application may include: a grouping processing module 11, an evaluation module 12 and a determination module 13, wherein:

[0184] The grouping processing module 11 is used to group the new citizens to be evaluated according to the user portrait data of the new citizens to be evaluated, and determine the group category to which the new citizens to be evaluated belong;

[0185] An evaluation module 12 is used to perform a credit rating evaluation based on the user portrait data and the credit rating evaluation model of the grouping category to obtain a preliminary credit rating quantitative value;

[0186] The determination module 13 is used to determine the target credit rating quantitative value of the new citizen to be evaluated based on the preliminary credit rating quantitative value.

[0187] The credit rating assessment device provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned credit rating assessment method embodiment of the present application. Its implementation principle and technical effect are similar and will not be repeated here.

[0188] In one embodiment, the grouping processing module 11 includes: a standardization processing unit and a grouping unit, wherein:

[0189] A standardization processing unit, used to perform standardization processing on the user portrait data to obtain standardized data;

[0190] The clustering unit is used to input the standardized data into a pre-trained clustering model to obtain the clustering category to which the new citizen to be evaluated belongs.

[0191] The credit rating assessment device provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned credit rating assessment method embodiment of the present application. Its implementation principle and technical effect are similar and will not be repeated here.

[0192] In one embodiment, the user portrait data includes numerical variable data and categorical variable data; the standardization processing unit includes: a processing subunit and a determination subunit, wherein:

[0193] A processing subunit is used to perform normalization processing on the numerical variable data to obtain normalized data; and to perform encoding processing on the categorical variable data to obtain encoded data;

[0194] A subunit is determined for determining the normalized data and the encoded data as normalized data.

[0195] The credit rating assessment device provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned credit rating assessment method embodiment of the present application. Its implementation principle and technical effect are similar and will not be repeated here.

[0196] In one embodiment, the credit rating assessment device further includes: a training data acquisition module, an analysis and processing module, a screening module and a clustering model training module, wherein:

[0197] A training data acquisition module is used to acquire training data of multiple sample new citizens;

[0198] The analysis and processing module is used to perform principal component analysis on the training data to obtain analysis results;

[0199] A screening module, used for screening first valid data from the training data according to the analysis result;

[0200] The clustering model training module is used to train the initial clustering model according to the first valid data to generate a clustering model.

[0201] The credit rating assessment device provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned credit rating assessment method embodiment of the present application. Its implementation principle and technical effect are similar and will not be repeated here.

[0202] In one embodiment, the credit rating assessment device further includes: a variable selection module and an assessment model training module, wherein:

[0203] A variable selection module is used to select variables for the training data of multiple sample new citizens to obtain second valid data;

[0204] The evaluation model training module is used to train the initial evaluation model according to the second valid data to obtain a credit rating evaluation model.

[0205] The credit rating assessment device provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned credit rating assessment method embodiment of the present application. Its implementation principle and technical effect are similar and will not be repeated here.

[0206] In one embodiment, the variable selection module includes: an encoding processing unit, a first determining unit, a second determining unit and a third determining unit, wherein:

[0207] A coding processing unit, used to perform evidence weight coding processing on each variable in the training data according to the training data to obtain converted data of each variable;

[0208] A first determining unit, used to determine the information value quantization value corresponding to each variable according to the converted data of each variable;

[0209] A second determining unit, configured to determine at least one valid variable among the variables according to the quantified value of the information value of each variable;

[0210] The third determining unit is used to obtain data of each valid variable from the training data, and determine the data of each valid variable as second valid data.

[0211] The credit rating assessment device provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned credit rating assessment method embodiment of the present application. Its implementation principle and technical effect are similar and will not be repeated here.

[0212] In one embodiment, the first determining unit is specifically configured to:

[0213] For any variable, the transformed data of the variable are grouped to obtain multiple data sets;

[0214] For any set of data sets, the data sets are converted into evidence weights to obtain the corresponding evidence weight values;

[0215] The variable value is evaluated according to the evidence weight value of each data set to determine the quantitative value of the information value corresponding to the variable.

[0216] The credit rating assessment device provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned credit rating assessment method embodiment of the present application. Its implementation principle and technical effect are similar and will not be repeated here.

[0217] In one embodiment, the determination module 13 is specifically configured to:

[0218] The preliminary credit rating quantified value is input into the comprehensive evaluation model of the credit rating, and the target credit rating quantified value is output; the comprehensive evaluation model of the credit rating is a model trained based on the sample quantified values ​​of the credit ratings of new users in different grouping categories.

[0219] The credit rating assessment device provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned credit rating assessment method embodiment of the present application. Its implementation principle and technical effect are similar and will not be repeated here.

[0220] For the specific definition of the credit rating assessment device, please refer to the definition of the credit rating assessment method above, which will not be repeated here. Each module in the above-mentioned credit rating assessment device can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0221] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Fig.10 As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide processing power. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store user portrait data and training data of new citizens to be evaluated. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a credit rating evaluation method is implemented.

[0222] Those skilled in the art will understand that Fig.10 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0223] In one embodiment, a computer device is also provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the technical solution in the above-mentioned credit rating assessment method embodiment of the present application is implemented. The implementation principle and technical effect are similar and will not be repeated here.

[0224] In one embodiment, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, the technical solution of the credit rating assessment method mentioned above in the present application is implemented. The implementation principle and technical effect are similar and will not be repeated here.

[0225] In one embodiment, a computer program product is also provided, including a computer program. When the computer program is executed by a processor, the technical solution of the credit rating assessment method described above in the present application is implemented. The implementation principle and technical effect are similar and will not be repeated here.

[0226] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0227] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0228] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A credit rating assessment method, characterized in that: The method comprises: According to the user portrait data of the new citizens to be evaluated, the new citizens to be evaluated are grouped and the group categories to which the new citizens to be evaluated belong are determined; Performing a credit rating assessment based on the user portrait data and the credit rating assessment model of the grouping category to obtain a preliminary credit rating quantitative value; The credit rating quantification value of the new citizen to be evaluated is determined according to the preliminary credit rating quantification value.

2. The method according to claim 1, characterized in that The step of grouping the new citizens to be evaluated based on the user portrait data of the new citizens to be evaluated and determining the group category to which the new citizens to be evaluated belong includes: Performing standardization processing on the user portrait data to obtain standardized data; The standardized data is input into a pre-trained clustering model to obtain the clustering category to which the new citizen to be evaluated belongs.

3. The method according to claim 2, characterized in that The user portrait data includes numerical variable data and categorical variable data; the step of performing standardization processing on the user portrait data to obtain standardized data includes: Normalizing the numerical variable data to obtain normalized data; and encoding the categorical variable data to obtain encoded data; The normalized data and the encoded data are determined as the standardized data.

4. The method according to any one of claims 1 to 3, characterized in that Before grouping the new citizens to be evaluated according to the user portrait data of the new citizens to be evaluated, the method further includes: Obtain training data of multiple sample new citizens; Performing principal component analysis on the training data to obtain analysis results; According to the analysis result, first valid data is selected from the training data; The initial clustering model is trained according to the first valid data to generate a clustering model.

5. The method according to any one of claims 1 to 3, characterized in that: The generation process of the credit rating assessment model includes: Perform variable selection on the training data of multiple sample new citizens to obtain the second effective data; The initial evaluation model is trained according to the second valid data to obtain the credit rating evaluation model.

6. The method according to claim 5, characterized in that The step of performing variable selection on the training data of multiple sample new citizens to obtain second valid data includes: According to the training data, each variable in the training data is subjected to evidence weight coding processing to obtain converted data of each variable; Determine the information value quantification value corresponding to each variable according to the converted data of each variable; Determining at least one valid variable among the variables according to the quantified value of the information value of each variable; The data of each of the valid variables is acquired from the training data, and the data of each of the valid variables is determined as the second valid data.

7. The method according to claim 6, characterized in that Determining the information value quantization value corresponding to each variable according to the converted data of each variable includes: For any variable, grouping the transformed data of the variable to obtain multiple data sets; For any set of data sets, performing evidence weight conversion processing on the data sets to obtain corresponding evidence weight values; The variable value is evaluated according to the evidence weight value of each data set to determine the quantitative value of the information value corresponding to the variable.

8. The method according to any one of claims 1 to 3, characterized in that Determining the target credit rating quantitative value of the new citizen to be evaluated according to the preliminary credit rating quantitative value includes: The preliminary credit rating quantized value is input into a comprehensive evaluation model of the credit rating, and the target credit rating quantized value is output; the comprehensive evaluation model of the credit rating is a model trained based on sample quantized values ​​of the credit ratings of new users of different grouping categories.

9. A credit rating assessment device, characterized in that: The device comprises: A grouping processing module, used to group the new citizens to be evaluated according to their user portrait data, and determine the grouping category to which the new citizens to be evaluated belong; An evaluation module, used to perform a credit rating evaluation based on the user portrait data and the credit rating evaluation model of the belonging group category to obtain a preliminary credit rating quantitative value; The determination module is used to determine the target credit rating quantitative value of the new citizen to be evaluated according to the preliminary credit rating quantitative value.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images