Credit risk assessment method and device, electronic equipment and storage medium

Abstract

The invention discloses a credit risk assessment method and device, electronic equipment and a storage medium. The method relates to the field of artificial intelligence. The method comprises the following steps: determining a target customer to be subjected to credit risk assessment; determining target data information of the target customer; the target data information is converted into a plurality of tags corresponding to the target customer based on a target tag library, the target tag library comprises a plurality of target interval tag sets, and each target interval tag set comprises a plurality of target interval tags of one variable; and inputting the plurality of tags into the target credit risk assessment model, and outputting an assessment result of the credit risk of the target customer. According to the method and the device, the problem that the accuracy of the evaluation result of the personal credit risk is low due to the fact that the processing of the personal credit data is not meticulous enough in the related technology is solved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and more specifically, to a credit risk assessment method, device, electronic device and storage medium. Background Art

[0002] Personal credit risk assessment is a key link in the daily operation of financial institutions. Its purpose is to predict whether the borrower can fulfill the repayment obligation on time, so as to help financial institutions make reasonable loan decisions and avoid potential economic losses. Traditional personal credit risk assessment methods mainly rely on fixed rules and expert experience, and usually can only process numerical data, such as the borrower's age, income, loan amount, etc. However, with the increasing complexity and personalization of financial services, the role of non-numerical data and uncertainty information in credit risk assessment is becoming increasingly prominent. This type of information includes the borrower's credit report description, detailed description of the loan purpose, and stability evaluation of employment status. They often contain rich qualitative knowledge and semantic information, which is crucial for a comprehensive assessment of the borrower's credit risk.

[0003] Although non-numerical data and uncertainty information are of great significance in personal credit risk assessment, existing technologies have obvious deficiencies in processing this data. Traditional risk assessment models are often based on statistical methods and threshold judgments, and have limited processing capabilities for text variables and fuzzy boundary data.

[0004] Currently, no effective solution has been proposed to address the problem that the relevant technologies do not process personal credit data in a detailed enough manner, resulting in low accuracy in the assessment results of personal credit risk. Summary of the invention

[0005] The main purpose of this application is to provide a credit risk assessment method, device, electronic device and storage medium to solve the problem that the processing of personal credit data in the relevant technology is not detailed enough, resulting in low accuracy of the assessment results of personal credit risk.

[0006] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a credit risk assessment method is provided. The method comprises: determining a target customer to be assessed for credit risk; determining target data information of the target customer; converting the target data information into a plurality of labels corresponding to the target customer based on a target label library, wherein the target label library comprises a plurality of target interval label sets, each target interval label set comprises a plurality of target interval labels of a variable; inputting the plurality of labels into a target credit risk assessment model, and outputting the assessment result of the credit risk of the target customer.

[0007] Furthermore, before converting the target data information into multiple labels corresponding to target customers based on the target label library, it includes: determining a training data set, wherein the training data set includes multiple variables; determining an interval label set corresponding to each variable in the training data set; based on the interval label set corresponding to each variable, calculating the initial entropy of each variable in the training data set; adjusting each interval label set based on the initial entropy of each variable to obtain multiple adjusted interval label sets; and constructing a target label library based on the multiple adjusted interval label sets.

[0008] Furthermore, each interval label set is adjusted based on the initial entropy of each variable to obtain multiple adjusted interval label sets, including: dividing the data of each variable in the training data set into N non-overlapping data intervals, and assigning an interval label to each data interval, where N is a positive integer greater than 1; calculating the fuzzy entropy of each data interval of each variable; based on the initial entropy of each variable and the fuzzy entropy of each data interval of each variable, adjusting N until a preset condition is met to obtain multiple adjusted interval label sets.

[0009] Furthermore, dividing the data of each variable in the training data set into N non-overlapping data intervals includes: if the variable in the training data set is a numeric variable, determining the numerical range and the number of intervals of the data of the numeric variable; based on the numerical range and the number of intervals of the data of the numeric variable, dividing the data of the numeric variable into N non-overlapping data intervals of equal proportions.

[0010] Furthermore, dividing the data of each variable in the training data set into N non-overlapping data intervals includes: if the variable in the training data set is a text variable, preprocessing the data of the text variable to obtain preprocessed data of the text variable; extracting features of the preprocessed data of the text variable; and clustering the preprocessed data of the text variable based on the features of the preprocessed data of the text variable to obtain N non-overlapping intervals.

[0011] Furthermore, calculating the fuzzy entropy of each data interval of each variable includes: calculating the membership interval of each data of each variable according to the membership function, wherein the membership interval is an interval between 0 and 1, and the membership is used to describe the degree to which each data belongs to the data interval; calculating the fuzzy entropy of each data interval of each variable according to the interval minimum value and interval maximum value of the membership interval of each data of each variable.

[0012] Furthermore, according to the membership function, calculating the membership interval of each data of each variable includes: determining the data interval to which each data of each variable belongs; determining the relative position of each data and the midpoint of the corresponding data interval; and calculating the membership interval of each data according to the relative position of each data and the midpoint of the corresponding data interval.

[0013] Furthermore, before inputting multiple labels into the target credit risk assessment model, the method also includes: determining the interval in which each data in the training data set is located according to the target label library; based on each data in the training data set and the interval in which each data is located, training the credit risk assessment model through the k-nearest neighbor algorithm to obtain the target credit risk assessment model.

[0014] In order to achieve the above-mentioned purpose, according to another aspect of the present application, a credit risk assessment device is provided. The device includes: a first determination unit, used to determine the target customer to be assessed for credit risk; a second determination unit, used to determine the target data information of the target customer; a conversion unit, used to convert the target data information into multiple labels corresponding to the target customer based on the target label library, wherein the target label library includes multiple target interval label sets, each target interval label set includes multiple target interval labels of a variable; an output unit, used to input the multiple labels into a target credit risk assessment model, and output the assessment result of the credit risk of the target customer.

[0015] Furthermore, the device also includes: a third determination unit, used to determine a training data set before converting the target data information into multiple labels corresponding to the target customers based on the target label library, wherein the training data set includes multiple variables; a fourth determination unit, used to determine the interval label set corresponding to each variable in the training data set; a first calculation unit, used to calculate the initial entropy of each variable in the training data set based on the interval label set corresponding to each variable; a first adjustment unit, used to adjust each interval label set based on the initial entropy of each variable to obtain multiple adjusted interval label sets; a construction unit, used to construct a target label library based on the multiple adjusted interval label sets.

[0016] Furthermore, the adjustment unit includes: a division unit, used to divide the data of each variable in the training data set into N non-overlapping data intervals, and assign an interval label to each data interval, wherein N is a positive integer greater than 1; a second calculation unit, used to calculate the fuzzy entropy of each data interval of each variable; a second adjustment unit, used to adjust N based on the initial entropy of each variable and the fuzzy entropy of each data interval of each variable until a preset condition is met, thereby obtaining a plurality of adjusted interval label sets.

[0017] Furthermore, the division unit includes: a determination module, which is used to determine the numerical range and number of intervals of the data of the numerical variable if the variable in the training data set is a numerical variable; and a division module, which is used to divide the data of the numerical variable into N non-overlapping data intervals of equal proportions based on the numerical range and number of intervals of the data of the numerical variable.

[0018] Furthermore, the division unit includes: a preprocessing module, which is used to preprocess the data of the text-type variable if the variable in the training data set is a text-type variable, so as to obtain the preprocessed data of the text-type variable; an extraction module, which is used to extract the features of the preprocessed data of the text-type variable; and a clustering module, which is used to cluster the preprocessed data of the text-type variable based on the features of the preprocessed data of the text-type variable, so as to obtain N non-overlapping intervals.

[0019] Furthermore, the second calculation unit includes: a first calculation module, used to calculate the membership interval of each data of each variable according to the membership function, wherein the membership interval is an interval between 0 and 1, and the membership is used to describe the degree to which each data belongs to the data interval; a second calculation module, used to calculate the fuzzy entropy of each data interval of each variable according to the interval minimum value and interval maximum value of the membership interval of each data of each variable.

[0020] Furthermore, the first calculation module includes: a first determination submodule, used to determine the data interval to which each data of each variable belongs; a second determination submodule, used to determine the relative position of each data and the midpoint of the corresponding data interval; and a calculation submodule, used to calculate the membership interval of each data based on the relative position of each data and the midpoint of the corresponding data interval.

[0021] Furthermore, the device also includes: a fifth determination unit, which is used to determine the interval of each data in the training data set according to the target label library before inputting multiple labels into the target credit risk assessment model; a training unit, which is used to train the credit risk assessment model through the k-nearest neighbor algorithm based on each data in the training data set and the interval of each data to obtain the target credit risk assessment model.

[0022] According to another aspect of the present application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein when the program is executed, a device where the computer-readable storage medium is located is controlled to execute any credit risk assessment method.

[0023] According to another aspect of the present application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include methods for executing any one of the credit risk assessment methods.

[0024] According to another aspect of the present application, a computer program product is provided, comprising computer instructions, which implement the steps of any one of the above-mentioned credit risk assessment methods when executed by a processor.

[0025] In an embodiment of the present application, by determining the target customer to be evaluated for credit risk; determining the target data information of the target customer; converting the target data information into multiple labels corresponding to the target customer based on the target label library, wherein the target label library includes multiple target interval label sets, each target interval label set includes multiple target interval labels of a variable; inputting multiple labels into the target credit risk assessment model, and outputting the assessment result of the credit risk of the target customer, the technical problem that the processing of personal credit data is not detailed enough, resulting in low accuracy of the assessment result of personal credit risk, is solved. In the present application, by obtaining the target data information of the target customer to be evaluated for credit risk, using the target label library to convert the target data information into multiple corresponding labels, and inputting multiple labels into the target credit risk assessment model, the assessment result of the credit risk of the target customer can be output, thereby achieving the technical effect of improving the accuracy of the credit risk assessment result. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0027] Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing a credit risk assessment method is shown;

[0028] Figure 2 is a flow chart of a credit risk assessment method provided according to an embodiment of the present application;

[0029] Figure 3 This is a flowchart of constructing a target tag library according to an embodiment of the present application;

[0030] Figure 4 is a schematic diagram of a credit risk assessment device provided according to an embodiment of the present application;

[0031] Figure 5 It is a structural block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0033] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application 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 interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0034] It should be noted that the relevant information (including but not limited to user device information, customer target data information, etc.) and data (including but not limited to training data sets, target label libraries, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. For example, an interface is set up between this system and the relevant users or organizations. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or organization through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or organization.

[0035] Example 1

[0036] According to an embodiment of the present application, a method embodiment for assessing credit risk is also provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0037] The credit risk assessment method embodiment provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 The hardware structure block diagram of a computer terminal (or mobile device) for implementing a credit risk assessment method is shown. Figure 1As shown, the computer terminal 10 (or mobile device) may include one or more (102a, 102b, ..., 102n are used to illustrate) processors 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0038] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10 (or mobile device). As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0039] The memory 104 can be used to store software programs and modules of application software, such as program instructions / data storage devices corresponding to the credit risk assessment method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, the credit risk assessment method described above is realized. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0040] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0041] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 (or mobile device).

[0042] Under the above operating environment, this application provides Figure 2 The credit risk assessment method shown. Figure 2 It is a flow chart of a credit risk assessment method provided according to an embodiment of the present application.

[0043] Step S201, determining target customers for credit risk assessment.

[0044] Optionally, in the credit business of a financial institution, when an individual or enterprise applies for a loan, a key decision faced by the financial institution is to assess the credit risk of the applicant. In this step, the financial institution will identify the individual or enterprise customers (the target customers mentioned above) who need to undergo credit risk assessment. For example, when someone submits an application for a personal loan online, the financial institution's system will identify them as a target customer for credit risk assessment.

[0045] Step S202, determining target data information of target customers.

[0046] Optionally, the target data information of the above-mentioned target customers may specifically include: age, gender, region, application amount, interest, loan term, loan purpose, employment status, income status, whether there is a loan record, and whether there is a default.

[0047] Step S203: converting the target data information into a plurality of labels corresponding to the target customers based on the target label library, wherein the target label library includes a plurality of target interval label sets, and each target interval label set includes a plurality of target interval labels of a variable.

[0048] Optionally, after receiving the target data information, the target label library can be used to convert the numerical and text data into a series of labels. For example, the income level is divided into three interval labels: "low income", "middle income", and "high income", so a customer with an annual income of 40,000 yuan may be labeled as having the characteristics of "middle income".

[0049] Step S204: input multiple tags into the target credit risk assessment model, and output the assessment result of the credit risk of the target customer.

[0050] Optionally, the labels obtained in step S203 are input into a trained target credit risk assessment model, which can be built based on a KNN algorithm or other supervised learning methods. The target credit risk assessment model predicts the credit risk level of the target customer based on the input labels, such as "low risk", "medium risk" or "high risk". For example, given a series of labels for a customer, such as "medium income" and "first loan", the target credit risk assessment model may predict that the customer is "medium risk".

[0051] For example, a financial institution needs to assess the credit risk of a new loan applicant, whose data are age: 27 years old; gender: male; income: 40,000 yuan / year; loan purpose: purchase of household appliances; historical loan record: none. The financial institution first determines that the applicant is the target customer to be evaluated (step S201). Next, the above collected data is used as target data information (step S202). Then, according to the target label library established previously, the income of "40,000 yuan / year" is converted into a "middle income" label, and "purchase of household appliances" is converted into a "consumer loan" label. Since gender and age are clear numerical values, they are directly converted into corresponding labels (step S203). Finally, these labels are input into the target credit risk assessment model, and the target credit risk assessment model outputs the credit risk level of the applicant, such as "medium risk" (step S204).

[0052] Through the above steps, financial institutions can convert target data information into multiple labels corresponding to target customers based on detailed information of target customers through the target label library, thereby efficiently and accurately assessing the credit risk of personal credit, providing a powerful tool for credit risk management of financial institutions.

[0053] In summary, the credit risk assessment method provided by the embodiment of the present application determines the target customer to be assessed for credit risk; determines the target data information of the target customer; converts the target data information into multiple labels corresponding to the target customer based on the target label library, wherein the target label library includes multiple target interval label sets, and each target interval label set includes multiple target interval labels of a variable; inputs the multiple labels into the target credit risk assessment model, and outputs the assessment result of the credit risk of the target customer, thereby solving the technical problem that the processing of personal credit data is not detailed enough, resulting in low accuracy of the assessment result of personal credit risk. In the present application, by obtaining the target data information of the target customer to be assessed for credit risk, using the target label library to convert the target data information into the corresponding multiple labels, and inputting the multiple labels into the target credit risk assessment model, the assessment result of the credit risk of the target customer can be output, thereby achieving the technical effect of improving the accuracy of the credit risk assessment result.

[0054] In an optional embodiment, Figure 3 is a flowchart of constructing a target tag library according to an embodiment of the present application, such as Figure 3 As shown, before converting the target data information into multiple labels corresponding to the target customers based on the target label library, it includes:

[0055] Step S301, determining a training data set, wherein the training data set includes multiple variables.

[0056] Optionally, the training data set may include data of historical personal credit customers, and the variables involved in the data set include age, gender, region, application amount, interest, loan term, loan purpose, employment status, income status, whether there is a loan record, and whether there is a default. For example, the determined training data set includes 30,000 data, which can be split into a training set and a test set at a ratio of 9:1.

[0057] Step S302, determining the interval label set corresponding to each variable in the training data set.

[0058] Optionally, a series of interval label sets need to be defined for each variable to fuzzify numerical or text data. For example, for the income variable, the interval label set can be defined as "low income", "middle income", and "high income", and the corresponding intervals can be set, such as annual income of 10,000-50,000 yuan is "low income", 50,001-1,296,011 yuan is "middle income", and more than 1,296,011 yuan is "high income".

[0059] For example, the variable "Purpose of borrowing" is divided into corresponding interval label sets. In financial institutions, customers have a variety of reasons for borrowing, including but not limited to education, medical treatment, travel, start-up capital, home decoration, etc. These reasons may be difficult to quantify in traditional models, but through fuzzy set theory, they can be converted into several fuzzy language labels. The reasons for borrowing can be divided into the following three target interval label sets:

[0060] Necessities: including education, medical care, emergency living expenses, etc. These reasons are closely related to the customer's basic living needs;

[0061] Leisure consumption: including travel, shopping, entertainment, etc. These reasons are more related to the customer's quality of life or personal preferences;

[0062] Investment and development: including start-up capital, home renovation, car and house purchase, etc. These reasons are related to the customer's investment behavior or long-term development plan;

[0063] For a customer whose loan purpose is "home improvement", he or she can be marked with the interval label of "investment and development".

[0064] For example, the interval label set corresponding to the division of the variable "age". Age is another important factor in credit risk assessment. Although age is a continuous numerical variable, it can be fuzzy processed through interval division to better reflect the impact of age on risk. For example, age can be divided into the following four target interval label sets:

[0065] Young applicants: 20-30 years old, may lack financial stability and experience;

[0066] Young and middle-aged applicants: 31-45 years old, usually with relatively stable financial status and a certain credit record;

[0067] Middle-aged applicants: 46-60 years old, likely to have a stable source of income and a good credit history;

[0068] Senior applicants: 61 years and older who may be facing financial uncertainty due to health or retirement.

[0069] For a customer who is 35 years old, he or she may be labeled as a "young and middle-aged applicant".

[0070] Step S303, based on the interval label set corresponding to each variable, calculate the initial entropy of each variable in the training data set.

[0071] Optionally, the calculation of the initial entropy can evaluate the uncertainty and information contained in each interval label set under the current partition.

[0072] Step S304: adjusting each interval label set based on the initial entropy of each variable to obtain a plurality of adjusted interval label sets.

[0073] Optionally, if it is found in step S303 that the initial entropy value of some variables is high, it means that the current interval division may not be reasonable or the amount of information is insufficient. The interval boundary can be adjusted or the number of labels can be increased to reduce the entropy value and improve the accuracy of information expression. For example, it is found that the entropy value of the "low income" label is high, so the interval can be further subdivided, such as dividing the interval of 10,000-50,000 yuan into "extremely low income" (10,000-25,000 yuan) and "lower income" (25,001-50,000 yuan). By iterative adjustment and optimizing the division of the interval label set, the expression ability and discrimination ability of the model training data can be improved.

[0074] Step S305 : constructing a target label library according to the multiple adjusted interval label sets.

[0075] Optionally, after completing the adjustment of the interval label set, these adjusted interval label sets can be integrated into a target label library for subsequent model training. An optimized target label library is constructed to more accurately reflect the distribution and complexity of customer data and improve the accuracy and robustness of credit risk assessment.

[0076] Through the above steps, financial institutions can build a target label library that accurately reflects data distribution and information expression capabilities. Based on this target label library, the target user's data information is converted into labels for risk assessment, which can improve the accuracy and efficiency of credit risk assessment and ensure the safety and sound development of the credit business of financial institutions.

[0077] In an optional embodiment, each interval label set is adjusted based on the initial entropy of each variable, and the multiple adjusted interval label sets are obtained, including:

[0078] The data of each variable in the training data set is divided into N non-overlapping data intervals, and an interval label is assigned to each data interval, where N is a positive integer greater than 1.

[0079] Optionally, you first need to divide each variable in the training data set into intervals to create an initial set of interval labels. For example, for the variable "income", its data intervals can be divided into: "extremely low income": 0-20,000 yuan / year; "low income": 20,001-40,000 yuan / year; "middle income": 40,001-80,000 yuan / year; "high income": 80,001-120,000 yuan / year; "extremely high income": more than 120,001 yuan / year.

[0080] Calculate the fuzzy entropy for each data interval of each variable.

[0081] Optionally, fuzzy entropy reflects the amount of information and degree of uncertainty of an interval label in distinguishing different output categories (such as credit risk level). The calculation of fuzzy entropy can use the interval value fuzzy entropy method to evaluate the contribution of each label to predicting credit risk. The calculation of fuzzy entropy provides a quantitative indicator for evaluating the effectiveness of interval label sets, which helps to identify and optimize those labels with insufficient distinguishing ability, thereby improving the predictive ability of the model.

[0082] For example, for the "middle income" interval label of the "income" variable, the fuzzy entropy of the data in this interval can be calculated when distinguishing different credit risk levels. If the entropy value is high, it means that the label's ability to distinguish is weak and needs further optimization.

[0083] Based on the initial entropy of each variable and the fuzzy entropy of each data interval of each variable, N is adjusted until the preset conditions are met, and a plurality of adjusted interval label sets are obtained.

[0084] For example, if the entropy value of the "low income" interval of the "income" variable is found to be high, it may be necessary to further subdivide it into "very low income" and "lower income" to improve the discrimination of information. Through repeated iterations, until the entropy value is reduced to a certain threshold or the model performance reaches the expected level, the adjusted interval label set is obtained.

[0085] For example, when processing the "income" variable, the income was initially divided into three intervals: "low income", "middle income" and "high income". However, after calculating the fuzzy entropy, it was found that the entropy value of the "middle income" interval was high, indicating that the current interval division may not be able to effectively distinguish the credit risk level. Therefore, the "middle income" interval was further subdivided into "lower income" (40,001-60,000 yuan / year) and "higher income" (60,001-80,000 yuan / year), and the entropy value was recalculated. After multiple adjustments and iterations, the five interval label sets of "extremely low income", "lower income", "middle income", "higher income" and "extremely high income" were finally determined. The entropy value at this time met the preset conditions, indicating that each label has good information content and discrimination ability when distinguishing credit risk levels.

[0086] In summary, through the above steps, financial institutions can adjust and optimize the interval label set to improve the accuracy and transparency of personal credit risk assessment, thereby strengthening their risk management capabilities.

[0087] In an optional embodiment, dividing the data of each variable in the training data set into N non-overlapping data intervals includes:

[0088] The first step is to determine the numerical range and number of intervals of the data of the numerical variable if the variables in the training data set are numerical variables.

[0089] Optionally, you first need to identify numeric variables, which are usually continuous values, such as customer age, loan amount, income, etc. Next, determine the value range of the numeric variable, that is, the minimum and maximum values ​​of the variable value. Then, based on business needs and data characteristics, select a number of intervals N, which must be a positive integer greater than 1, representing the number of intervals into which the value range is divided.

[0090] For example, when processing the numerical variable "income", it can be found that the minimum income in the training data set is 10,000 yuan / year and the maximum income is 3,296,011 yuan / year. The income data can be divided into 5 intervals (N=5).

[0091] The second step is to divide the data of the numerical variable into N non-overlapping data intervals of equal proportions based on the numerical range and number of intervals of the data of the numerical variable.

[0092] Optionally, after determining the range and number of intervals, the data of the numeric variable can be evenly divided into N non-overlapping intervals. This is usually achieved by calculating the width of each interval, that is, (maximum value - minimum value) / N. Each interval should be assigned a unique label to represent the data characteristics within the interval. By dividing the intervals in equal proportions, the homogeneity of the data within each interval label and the mutual exclusivity between intervals are ensured, reducing the deviation caused by uneven data segmentation.

[0093] For example, taking the variable "income" as an example, it is found that the income range is 10,000 yuan / year to 3,296,011 yuan / year, N=5, then the width of each interval is (3,296,011-10,000) / 5=58,000 yuan. In this way, the income data can be divided into the following five intervals:

[0094] The first interval: 10,000 yuan / year to 68,000 yuan / year, labeled as "extremely low income";

[0095] The second interval: 68,001 yuan / year to 126,000 yuan / year, labeled as "low income";

[0096] The third interval: 126,001 yuan / year to 184,000 yuan / year, labeled as "middle income";

[0097] The fourth interval: 184,001 yuan / year to 242,000 yuan / year, labeled as "higher income";

[0098] The fifth interval: 242,001 yuan / year to 3,296,011 yuan / year, labeled as “extremely high income”.

[0099] In summary, through the above steps, financial institutions can convert the data of numerical variables into a series of interval labels that are easy to understand and process, providing a solid data foundation for credit risk assessment, improving prediction accuracy, and thus strengthening their risk management capabilities. This data preprocessing method is particularly suitable for processing complex numerical data that is difficult to effectively use in traditional rule models, and can better capture fuzzy and uncertain information in the data, providing strong support for the credit decision-making of financial institutions.

[0100] In an optional embodiment, dividing the data of each variable in the training data set into N non-overlapping data intervals includes:

[0101] If the variables in the training data set are text variables, the data of the text variables are preprocessed to obtain the preprocessed data of the text variables.

[0102] Optionally, the above preprocessing is a basic step for analyzing the data of text variables, mainly used to clean and standardize the original text data for subsequent feature extraction and analysis. Preprocessing can include removing stop words, punctuation marks, numbers and other non-key words, converting to lowercase, standardizing phrases and synonyms, and possible stemming or word form restoration.

[0103] For example, when processing the text variable "Purpose of borrowing", you will encounter descriptions such as "education", "home improvement", "medical care", "entrepreneurship", etc. The purpose of preprocessing is to convert these text descriptions into a processable form.

[0104] Extract the features of the preprocessed text variable data.

[0105] Optionally, feature extraction is a key step in text analysis, and its goal is to convert the preprocessed text into a series of numerical features. Feature extraction can be achieved through methods such as bag-of-words model, TF-IDF (term frequency-inverse document frequency), word embedding, etc.

[0106] For example, the TF-IDF method can be used to extract features of the data of the text variable "Purpose of borrowing". A vocabulary can be constructed and a TF-IDF value can be calculated for each word to indicate its importance in the text. For example, "education" may appear frequently in multiple texts, but considering its prevalence in all documents, the TF-IDF method will reduce its weight, while "entrepreneurship" may appear less frequently, but its uniqueness in the document makes its TF-IDF value higher, thus being regarded as a more important feature.

[0107] The preprocessed text variable data is clustered based on the features of the preprocessed text variable data to obtain N non-overlapping intervals.

[0108] Optionally, cluster analysis is an unsupervised learning method used to group data points with similar characteristics into disjoint intervals or "clusters". For text data, clustering can help identify groups of customers with similar loan purposes, which can then create a more meaningful set of interval labels.

[0109] For example, for the text variable "Purpose of loan", clustering algorithms such as K-means, hierarchical clustering or DBSCAN can be used to cluster text data based on TF-IDF features. Assuming N = 4, 4 clusters may be obtained, each cluster representing a group of customers with similar loan purposes, such as "education", "family", "medical care" and "entrepreneurship".

[0110] In summary, through the above steps, text variables that are difficult to process directly can be converted into a series of interval label sets, which significantly improves the processing ability and prediction accuracy of text data. This processing method is particularly suitable for processing text information that is difficult to use in traditional models, such as loan purpose, employment status, etc. It can better capture the semantic information and customer intentions in the data, and provide more accurate support for the credit decision-making of financial institutions.

[0111] In an optional embodiment, calculating the fuzzy entropy of each data interval of each variable includes:

[0112] According to the membership function, the membership interval of each data of each variable is calculated, wherein the membership interval is an interval between 0 and 1, and the membership is used to describe the degree to which each data belongs to the data interval.

[0113] Optionally, a membership function can describe the degree to which a data point belongs to a set of intervals. One or more membership functions can be defined for each variable, which can be triangular, trapezoidal, or Gaussian, converting numerical or textual features into a membership interval between 0 and 1. This interval reflects the range of likelihood that a data point belongs to a particular interval or class, thereby quantifying uncertainty.

[0114] For example, when dealing with the variable "income", a trapezoidal membership function can be defined to describe the fuzzy set "medium income". Assuming that the boundaries of the trapezoidal membership function of "medium income" are (30000, 50000, 70000, 90000), then for a customer with an income of 55,000 yuan, according to the trapezoidal membership function, the membership interval of his "medium income" is [0.5, 1], because 55,000 falls within the hypotenuse interval, and the minimum and maximum values ​​of the hypotenuse interval correspond to the values ​​of the upper left and upper right corners of the trapezoid, respectively.

[0115] According to the interval minimum value and interval maximum value of the membership interval of each data of each variable, the fuzzy entropy of each data interval of each variable is calculated.

[0116] Optionally, fuzzy entropy is an indicator that quantifies the uncertainty of a fuzzy set, which is calculated based on the membership interval of the data point. Specifically, the calculation of fuzzy entropy depends on the interval minimum and interval maximum of the membership interval, which describe the possible range of a data point belonging to a certain category. For each data interval, the calculation of fuzzy entropy can reflect the uncertainty and information content of the data in the interval when distinguishing different loan risk categories.

[0117] For example, the membership interval of "tuition payment" in the "education" set is [0.8, 1], indicating that the loan application is highly likely to belong to the "education" category, while the membership interval in the "entrepreneurship" set is [0, 0.1], indicating that it is almost impossible to belong to the "entrepreneurship" category. Next, the fuzzy entropy is calculated based on the interval minimum and interval maximum of each membership interval. For "tuition payment", its fuzzy entropy in the "education" set is lower, while the fuzzy entropy in the "entrepreneurship" set is higher. This means that when "tuition payment" is classified into the "education" set, the model's assessment of loan risk is more certain, while in the "entrepreneurship" set there is a higher uncertainty.

[0118] In summary, the above steps can process information that is difficult to accurately quantify in traditional numerical models, such as description of loan purpose, employment status, etc., and can more comprehensively assess the credit risk of personal credit, thereby making more reasonable and accurate credit decisions. At the same time, the calculation of fuzzy entropy also enhances the transparency and interpretability of the model, which helps financial institutions understand the decision-making process of the model and improve the efficiency and effectiveness of risk management.

[0119] In an optional embodiment, calculating the membership interval of each data of each variable according to the membership function includes:

[0120] The first step is to determine the data interval to which each data of each variable belongs;

[0121] For example, the variable "loan term" is being processed and divided into three intervals: "short-term" (<1 year), "medium-term" (1-3 years) and "long-term" (>3 years). For an application with a loan term of 2.5 years, it will be classified into the data interval of "medium-term".

[0122] The second step is to determine the relative position of each data point and the midpoint of the corresponding data interval.

[0123] For example, if we continue to use the data point with a "loan term" of 2.5 years as an example, the midpoint of the medium-term loan range is 2 years (assuming the range is 1-3 years). Therefore, the relative position of 2.5 years to the midpoint of 2 years is 0.5 years, that is, the data point is 0.5 years above the midpoint.

[0124] The third step is to calculate the membership interval of each data according to the relative position of each data and the midpoint of the corresponding data interval.

[0125] For example, the variable "income" is divided into three interval labels: "low income", "middle income" and "high income". The "low income" set is defined as an annual income between 20,000 and 40,000 yuan. Among them, 20,000 yuan and 40,000 yuan are the boundaries of the set, and 30,000 yuan is the midpoint. The "middle income" set is defined as an annual income between 35,000 and 65,000 yuan, with a midpoint of 50,000 yuan. The "high income" set is defined as an annual income between 60,000 and 80,000 yuan, with a midpoint of 70,000 yuan. A customer has an annual income of 55,000 yuan, and the range of the "middle income" set includes 55,000 yuan, so this data point belongs to the "middle income" set. Next, calculate the relative position of 55,000 yuan relative to the midpoint of the "middle income" set, 50,000 yuan. 55,000 yuan is to the right of the midpoint of 50,000 yuan, 5,000 yuan away from the midpoint. Based on the relative position of 55,000 yuan and the midpoint 50,000 yuan and the triangular membership function, its membership interval is calculated. The membership of the triangular membership function at the midpoint is 1, and gradually decreases to 0 toward the two boundaries. For 55,000 yuan, it is 5,000 yuan away from the midpoint, and there is still a distance of 10,000 yuan from the boundary of the "middle-income" set of 65,000 yuan. According to the triangular membership function, the membership interval of 55,000 yuan is calculated as follows:

[0126] 55,000 yuan is to the right of the midpoint, 5,000 yuan from the midpoint and 10,000 yuan from the right boundary;

[0127] According to the triangular membership function, the proportion of membership reduction is the same for every 1 yuan away from the center in the direction from the center to the boundary;

[0128] The membership calculation of 55,000 yuan should take into account its distance from the midpoint and its position on the base of the triangle, that is, its position relative to the hypotenuse of the base of the triangle;

[0129] Since 55,000 yuan is 5,000 yuan away from the midpoint and 10,000 yuan away from the right boundary, its membership interval can be calculated using the slope of the triangular membership function;

[0130] Assuming the width of the base of the triangle is 30,000 yuan (from 35,000 yuan to 65,000 yuan), then the slope (the rate at which the degree of membership changes) is 1 / 15,000 (calculated from the width of 30,000 yuan and the width of the top of the triangle, 1);

[0131] Therefore, the membership interval of 55,000 yuan will start from its slope calculation from the midpoint of 5,000 yuan until the membership of 1 at the top of the triangle;

[0132] The calculated membership interval is [0.6667, 1] (here 0.6667 is obtained by dividing the distance of 5000 yuan by the width of the bottom of the triangle 30000 yuan, rounded to 0.6667).

[0133] In summary, by introducing the membership interval in the above steps, the specific risk status of each customer can be assessed more delicately, avoiding the information simplification caused by simple threshold division, and improving the accuracy and reliability of model decision-making.

[0134] In an optional embodiment, before inputting the multiple tags into the target credit risk assessment model, the method further includes:

[0135] The first step is to determine the interval of each data in the training data set according to the target label library;

[0136] Optionally, the interval corresponding to each data point in the training data set is determined, that is, the feature of each data point is mapped to a certain interval in the target label library, which can be used for subsequent model training.

[0137] In the second step, based on each data in the training data set and the interval in which each data is located, the credit risk assessment model is trained by the k-nearest neighbor algorithm to obtain the target credit risk assessment model.

[0138] Optionally, k-NN is an instance-based learning algorithm that predicts the category of a sample to be classified by calculating the similarity between the sample to be classified and the samples in the training set. In this step, the k-NN algorithm will be applied to the training data set to train the target credit risk assessment model. The credit risk assessment model will learn how to predict the credit risk label of a data point based on the characteristics of the data point based on the information in the interval where the training data point is located.

[0139] For example, use the k-NN algorithm to train a credit risk assessment model, and set k = 3. When evaluating a new loan application, the credit risk assessment model calculates the distance between this application and all applications in the training set (which can be Euclidean distance, Manhattan distance, etc.). Then, the credit risk assessment model will find the three closest training data points and vote based on the credit risk labels of these data points, and use the label with the most votes as the credit risk prediction of the new application. Suppose that in the training data set, there are three loan applications (A, B, C) that are closest to the new application, where the credit risk labels of applications A and B are "low risk" and the credit risk label of application C is "medium risk". Then, according to the k-NN algorithm, the credit risk of the new application will be predicted as "low risk".

[0140] In summary, a target credit risk assessment model can be trained based on each data point in the training data set and the k-NN algorithm. The target credit risk assessment model can not only handle the uncertainty and ambiguity in the data, but also learn how to predict the credit risk of personal loan applications based on the characteristics and interval attribution of each data point in the training data set.

[0141] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0142] Example 2

[0143] The embodiment of the present application also provides a credit risk assessment device. It should be noted that the credit risk assessment device of the embodiment of the present application can be used to execute the credit risk assessment method provided in the embodiment of the present application. The credit risk assessment device provided in the embodiment of the present application is introduced below.

[0144] According to an embodiment of the present application, a device for implementing the above credit risk assessment method is also provided. Figure 4 is a schematic diagram of a credit risk assessment device provided according to an embodiment of the present application, such as Figure 4 As shown, the device includes: a first determination unit 401, a second determination unit 402, a conversion unit 403, and an output unit 404.

[0145] Specifically, the first determination unit 401 determines the target customer to be evaluated for credit risk;

[0146] The second determining unit 402 determines target data information of the target customer;

[0147] The conversion unit 403 converts the target data information into multiple labels corresponding to the target customers based on the target label library, wherein the target label library includes multiple target interval label sets, and each target interval label set includes multiple target interval labels of a variable;

[0148] The output unit 404 inputs the multiple tags into the target credit risk assessment model and outputs the assessment result of the credit risk of the target customer.

[0149] The credit risk assessment device provided in the embodiment of the present application determines the target customer to be assessed for credit risk through the first determination unit 401; the second determination unit 402 determines the target data information of the target customer; the conversion unit 403 converts the target data information into multiple tags corresponding to the target customer based on the target tag library, wherein the target tag library includes multiple target interval tag sets, and each target interval tag set includes multiple target interval tags of a variable; the output unit 404 inputs multiple tags into the target credit risk assessment model, and outputs the assessment result of the credit risk of the target customer, which solves the technical problem that the processing of personal credit data is not detailed enough, resulting in low accuracy of the assessment result of personal credit risk. Thus, the technical effect of improving the accuracy of the assessment result of credit risk is achieved.

[0150] Optionally, in the credit risk assessment device provided in the embodiment of the present application, the device also includes: a third determination unit, used to determine a training data set before converting the target data information into multiple labels corresponding to target customers based on the target label library, wherein the training data set includes multiple variables; a fourth determination unit, used to determine the interval label set corresponding to each variable in the training data set; a first calculation unit, used to calculate the initial entropy of each variable in the training data set based on the interval label set corresponding to each variable; a first adjustment unit, used to adjust each interval label set based on the initial entropy of each variable to obtain multiple adjusted interval label sets; a construction unit, used to construct a target label library based on the multiple adjusted interval label sets.

[0151] Optionally, in the credit risk assessment device provided in the embodiment of the present application, the adjustment unit includes: a division unit, used to divide the data of each variable in the training data set into N non-overlapping data intervals, and assign an interval label to each data interval, where N is a positive integer greater than 1; a second calculation unit, used to calculate the fuzzy entropy of each data interval of each variable; a second adjustment unit, used to adjust N based on the initial entropy of each variable and the fuzzy entropy of each data interval of each variable until a preset condition is met, thereby obtaining multiple adjusted interval label sets.

[0152] Optionally, in the credit risk assessment device provided in the embodiment of the present application, the division unit includes: a determination module, used to determine the numerical range and the number of intervals of the data of the numerical variable if the variable in the training data set is a numerical variable; a division module, used to divide the data of the numerical variable into N non-overlapping data intervals of equal proportions based on the numerical range and the number of intervals of the data of the numerical variable.

[0153] Optionally, in the credit risk assessment device provided in the embodiment of the present application, the division unit includes: a preprocessing module, which is used to preprocess the data of the text-type variable if the variable in the training data set is a text-type variable, so as to obtain the preprocessed data of the text-type variable; an extraction module, which is used to extract the features of the preprocessed data of the text-type variable; and a clustering module, which is used to cluster the preprocessed data of the text-type variable based on the features of the preprocessed data of the text-type variable, so as to obtain N non-overlapping intervals.

[0154] Optionally, in the credit risk assessment device provided in the embodiment of the present application, the second calculation unit includes: a first calculation module, used to calculate the membership interval of each data of each variable according to the membership function, wherein the membership interval is an interval between 0 and 1, and the membership is used to describe the degree to which each data belongs to the data interval; a second calculation module, used to calculate the fuzzy entropy of each data interval of each variable according to the interval minimum value and interval maximum value of the membership interval of each data of each variable.

[0155] Optionally, in the credit risk assessment device provided in the embodiment of the present application, the first calculation module includes: a first determination submodule, used to determine the data interval to which each data of each variable belongs; a second determination submodule, used to determine the relative position of each data and the midpoint of the corresponding data interval; and a calculation submodule, used to calculate the membership interval of each data based on the relative position of each data and the midpoint of the corresponding data interval.

[0156] Optionally, in the credit risk assessment device provided in the embodiment of the present application, the device also includes: a fifth determination unit, used to determine the interval in which each data in the training data set is located according to the target label library before inputting multiple labels into the target credit risk assessment model; a training unit, used to train the credit risk assessment model through the k-nearest neighbor algorithm based on each data in the training data set and the interval in which each data is located to obtain the target credit risk assessment model.

[0157] It should be noted that the first determination unit 401, the second determination unit 402, the conversion unit 403, and the output unit 404 correspond to steps S201 to S204 in Example 1, and the two modules and the corresponding steps implement the same examples and application scenarios, but are not limited to the contents disclosed in the above-mentioned Example 1. It should be noted that the above-mentioned modules or units can be hardware components or software components stored in a memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n), and the above-mentioned modules can also be part of the device and can be run in the computer terminal 10 provided in Example 1.

[0158] Example 3

[0159] The embodiment of the present application may provide a computer terminal, which may be any computer terminal device in a computer terminal group. Optionally, in this embodiment, the computer terminal may also be replaced by a terminal device such as a mobile terminal or an electronic device.

[0160] Optionally, in this embodiment, the computer terminal may be located in at least one network device among a plurality of network devices of the computer network.

[0161] In this embodiment, the above-mentioned computer terminal can execute the program code of the following steps in the credit risk assessment method: determining the target customer to be assessed for credit risk; determining the target data information of the target customer; converting the target data information into multiple labels corresponding to the target customer based on the target label library, wherein the target label library includes multiple target interval label sets, each target interval label set includes multiple target interval labels of a variable; inputting the multiple labels into the target credit risk assessment model, and outputting the assessment result of the credit risk of the target customer.

[0162] Optionally, the computer terminal may execute the program code of the following steps in the credit risk assessment method: before converting the target data information into multiple labels corresponding to the target customers based on the target label library, including: determining a training data set, wherein the training data set includes multiple variables; determining an interval label set corresponding to each variable in the training data set; based on the interval label set corresponding to each variable, calculating the initial entropy of each variable in the training data set; adjusting each interval label set based on the initial entropy of each variable to obtain multiple adjusted interval label sets; and constructing a target label library based on the multiple adjusted interval label sets.

[0163] Optionally, the computer terminal may execute the program code of the following steps in the credit risk assessment method: adjusting each interval label set based on the initial entropy of each variable to obtain multiple adjusted interval label sets, including: dividing the data of each variable in the training data set into N non-overlapping data intervals, and assigning an interval label to each data interval, where N is a positive integer greater than 1; calculating the fuzzy entropy of each data interval of each variable; adjusting N based on the initial entropy of each variable and the fuzzy entropy of each data interval of each variable until a preset condition is met to obtain multiple adjusted interval label sets.

[0164] Optionally, the computer terminal may execute the program code of the following steps in the credit risk assessment method: dividing the data of each variable in the training data set into N non-overlapping data intervals, including: if the variable in the training data set is a numerical variable, determining the numerical range and number of intervals of the data of the numerical variable; based on the numerical range and number of intervals of the data of the numerical variable, dividing the data of the numerical variable into N non-overlapping data intervals of equal proportions.

[0165] Optionally, the computer terminal may execute the program code of the following steps in the credit risk assessment method: dividing the data of each variable in the training data set into N non-overlapping data intervals, including: if the variable in the training data set is a text variable, preprocessing the data of the text variable to obtain preprocessed data of the text variable; extracting features of the preprocessed data of the text variable; clustering the preprocessed data of the text variable based on the features of the preprocessed data of the text variable to obtain N non-overlapping intervals.

[0166] Optionally, the computer terminal can execute the following program codes in the credit risk assessment method: calculating the fuzzy entropy of each data interval of each variable includes: calculating the membership interval of each data of each variable according to the membership function, wherein the membership interval is an interval between 0 and 1, and the membership is used to describe the degree to which each data belongs to the data interval; calculating the fuzzy entropy of each data interval of each variable according to the interval minimum value and interval maximum value of the membership interval of each data of each variable.

[0167] Optionally, the computer terminal may execute the program code of the following steps in the credit risk assessment method: calculating the membership interval of each data of each variable according to the membership function, including: determining the data interval to which each data of each variable belongs; determining the relative position of each data and the midpoint of the corresponding data interval; calculating the membership interval of each data according to the relative position of each data and the midpoint of the corresponding data interval.

[0168] Optionally, the above-mentioned computer terminal can execute the program code of the following steps in the credit risk assessment method: before inputting multiple labels into the target credit risk assessment model, the method also includes: determining the interval in which each data in the training data set is located according to the target label library; based on each data in the training data set and the interval in which each data is located, training the credit risk assessment model through the k-nearest neighbor algorithm to obtain the target credit risk assessment model.

[0169] Optionally, Figure 5 is a structural block diagram of an electronic device according to an embodiment of the present application. Figure 5 As shown, the electronic device may include: one or more ( Figure 5 (only one is shown) processor 1002, memory 1004, storage controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0170] Among them, the memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the credit risk assessment method and device in the embodiment of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned credit risk assessment method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory may further include a memory remotely arranged relative to the processor, and these remote memories can be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0171] The processor may call the information and application programs stored in the memory through the transmission device to execute the above steps in the above credit risk assessment method.

[0172] By adopting the embodiment of the present application, a scheme for assessing credit risk is provided. By determining the target customer to be assessed for credit risk; determining the target data information of the target customer; converting the target data information into multiple labels corresponding to the target customer based on the target label library, wherein the target label library includes multiple target interval label sets, and each target interval label set includes multiple target interval labels of a variable; inputting the multiple labels into the target credit risk assessment model, and outputting the assessment result of the credit risk of the target customer, the technical problem that the processing of personal credit data is not detailed enough, resulting in low accuracy of the assessment result of personal credit risk, is solved. In the present application, by obtaining the target data information of the target customer to be assessed for credit risk, using the target label library to convert the target data information into the corresponding multiple labels, and inputting the multiple labels into the target credit risk assessment model, the assessment result of the credit risk of the target customer can be output, thereby achieving the technical effect of improving the accuracy of the assessment result of the credit risk.

[0173] It can be understood by those skilled in the art that Figure 5 The structure shown is for illustration only, and the electronic device may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile Internet device (Mobile Internet Devices, MID), a PAD, or other terminal devices. Figure 5 The structure of the electronic device is not limited. Figure 5 More or fewer components (such as network interfaces, display devices, etc.) shown in, or having Figure 5 Different configurations are shown.

[0174] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0175] Example 4

[0176] The embodiment of the present application further provides a storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the credit risk assessment method provided in the first embodiment.

[0177] Optionally, in this embodiment, the above storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0178] Optionally, in this embodiment, the storage medium is configured to store program codes for executing the following steps: determining a target customer to be evaluated for credit risk; determining target data information of the target customer; converting the target data information into multiple labels corresponding to the target customer based on a target label library, wherein the target label library includes multiple target interval label sets, each target interval label set includes multiple target interval labels for a variable; inputting the multiple labels into a target credit risk assessment model, and outputting an assessment result of the credit risk of the target customer.

[0179] Optionally, the storage medium is also configured to store program code for executing the following steps: before converting the target data information into multiple labels corresponding to target customers based on the target label library, including: determining a training data set, wherein the training data set includes multiple variables; determining an interval label set corresponding to each variable in the training data set; based on the interval label set corresponding to each variable, calculating the initial entropy of each variable in the training data set; adjusting each interval label set based on the initial entropy of each variable to obtain multiple adjusted interval label sets; and constructing a target label library based on the multiple adjusted interval label sets.

[0180] Optionally, the storage medium is also configured to store program code for executing the following steps: adjusting each interval label set based on the initial entropy of each variable to obtain multiple adjusted interval label sets, including: dividing the data of each variable in the training data set into N non-overlapping data intervals, and assigning an interval label to each data interval, where N is a positive integer greater than 1; calculating the fuzzy entropy of each data interval of each variable; adjusting N based on the initial entropy of each variable and the fuzzy entropy of each data interval of each variable until a preset condition is met to obtain multiple adjusted interval label sets.

[0181] Optionally, the storage medium is also configured to store program code for executing the following steps: dividing the data of each variable in the training data set into N non-overlapping data intervals, including: if the variable in the training data set is a numerical variable, determining the numerical range and number of intervals of the data of the numerical variable; based on the numerical range and number of intervals of the data of the numerical variable, dividing the data of the numerical variable into N non-overlapping data intervals of equal proportions.

[0182] Optionally, the storage medium is also configured to store program code for executing the following steps: dividing the data of each variable in the training data set into N non-overlapping data intervals including: if the variable in the training data set is a text variable, preprocessing the data of the text variable to obtain preprocessed data of the text variable; extracting features of the preprocessed data of the text variable; clustering the preprocessed data of the text variable based on the features of the preprocessed data of the text variable to obtain N non-overlapping intervals.

[0183] Optionally, the storage medium is also configured to store program code for executing the following steps: calculating the fuzzy entropy of each data interval of each variable includes: calculating the membership interval of each data of each variable according to the membership function, wherein the membership interval is an interval between 0 and 1, and the membership is used to describe the degree to which each data belongs to the data interval; calculating the fuzzy entropy of each data interval of each variable according to the interval minimum value and interval maximum value of the membership interval of each data of each variable.

[0184] Optionally, the storage medium is also configured to store program codes for executing the following steps: calculating the membership interval of each data of each variable according to the membership function, including: determining the data interval to which each data of each variable belongs; determining the relative position of each data and the midpoint of the corresponding data interval; and calculating the membership interval of each data according to the relative position of each data and the midpoint of the corresponding data interval.

[0185] Optionally, the storage medium is also configured to store program code for executing the following steps: before inputting multiple labels into the target credit risk assessment model, the method also includes: determining the interval in which each data in the training data set is located according to the target label library; based on each data in the training data set and the interval in which each data is located, training the credit risk assessment model through the k-nearest neighbor algorithm to obtain the target credit risk assessment model.

[0186] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing the program steps of the credit risk assessment method.

[0187] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0188] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0189] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0190] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0191] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0192] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk and other media that can store program codes.

[0193] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A credit risk assessment method, characterized in that: include: Identify target customers for credit risk assessment; Determining target data information of the target customers; Converting the target data information into a plurality of labels corresponding to the target customers based on a target label library, wherein the target label library includes a plurality of target interval label sets, and each target interval label set includes a plurality of target interval labels of a variable; The multiple tags are input into a target credit risk assessment model, and an assessment result of the credit risk of the target customer is output.

2. The method according to claim 1, characterized in that Before converting the target data information into a plurality of labels corresponding to the target customers based on the target label library, the method includes: Determine a training data set, wherein the training data set includes a plurality of variables; Determine an interval label set corresponding to each variable in the training data set; Calculate the initial entropy of each variable in the training data set based on the interval label set corresponding to each variable; Adjusting each interval label set based on the initial entropy of each variable to obtain a plurality of adjusted interval label sets; The target label library is constructed according to the multiple adjusted interval label sets.

3. The method according to claim 2, characterized in that Each interval label set is adjusted based on the initial entropy of each variable to obtain multiple adjusted interval label sets including: Divide the data of each variable in the training data set into N non-overlapping data intervals, and assign an interval label to each data interval, where N is a positive integer greater than 1; Calculate the fuzzy entropy of each data interval for each variable; Based on the initial entropy of each variable and the fuzzy entropy of each data interval of each variable, N is adjusted until a preset condition is met, thereby obtaining a plurality of adjusted interval label sets.

4. The method according to claim 3, characterized in that Dividing the data of each variable in the training data set into N disjoint data intervals includes: If the variables in the training data set are numerical variables, then determine the numerical range and number of intervals of the data of the numerical variables; Based on the numerical range and the number of intervals of the data of the numerical variable, the data of the numerical variable is divided into N non-overlapping data intervals of equal proportion.

5. The method according to claim 3, characterized in that: Dividing the data of each variable in the training data set into N disjoint data intervals includes: If the variables in the training data set are text variables, preprocessing the data of the text variables to obtain preprocessed data of the text variables; Extracting features of the preprocessed text variable data; The preprocessed text variable data is clustered based on the features of the preprocessed text variable data to obtain N non-overlapping intervals.

6. The method according to claim 3, characterized in that Calculating the fuzzy entropy of each data interval for each variable includes: According to the membership function, the membership interval of each data of each variable is calculated, wherein the membership interval is an interval between 0 and 1, and the membership is used to describe the degree to which each data belongs to the data interval; The fuzzy entropy of each data interval of each variable is calculated according to the interval minimum value and the interval maximum value of the membership interval of each data of each variable.

7. The method according to claim 6, characterized in that According to the membership function, the membership interval of each data of each variable is calculated including: Determine the data interval to which each data of each variable belongs; Determine the relative position of each data point to the midpoint of the corresponding data interval; According to the relative position of each data and the midpoint of the corresponding data interval, the membership interval of each data is calculated.

8. The method according to claim 2, characterized in that: Before inputting the multiple tags into the target credit risk assessment model, the method further includes: Determine the interval where each data in the training data set is located according to the target label library; Based on each data in the training data set and the interval in which each data is located, a credit risk assessment model is trained by a k-nearest neighbor algorithm to obtain a target credit risk assessment model.

9. A credit risk assessment device, characterized in that: include: A first determination unit is used to determine target customers to be evaluated for credit risk; A second determining unit, used to determine target data information of the target customer; A conversion unit, configured to convert the target data information into a plurality of labels corresponding to the target customers based on a target label library, wherein the target label library includes a plurality of target interval label sets, each target interval label set including a plurality of target interval labels of a variable; The output unit is used to input the multiple tags into the target credit risk assessment model and output the assessment result of the credit risk of the target customer.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored executable program, wherein when the executable program is executed, the device where the computer-readable storage medium is located is controlled to execute the credit risk assessment method according to any one of claims 1 to 8.

11. An electronic device, characterized in that: include: A memory storing an executable program; A processor is used to run the program, wherein the program, when running, executes the credit risk assessment method described in any one of claims 1 to 8.

12. A computer program product comprising computer instructions, characterized in that: When the computer instructions are executed by a processor, the steps of the credit risk assessment method described in any one of claims 1 to 8 are implemented.

Images