Early warning analysis method and device, equipment, storage medium and product

Abstract

The invention relates to the technical field of computers, and particularly provides an early warning analysis method and device, equipment, a storage medium and a product, and the method comprises the steps: carrying out the early warning analysis of a user according to at least one piece of user attribute data of the user, and obtaining a first credit degree and a first feature value of the user, each piece of user attribute data comprises attribute data corresponding to at least one user attribute of the user; a plurality of early warning areas in the coordinate system are obtained, the plurality of early warning areas are obtained through division based on an early warning curve in the coordinate system, the plurality of early warning areas are associated with corresponding early warning levels respectively, and the coordinate system is obtained through construction with the credits and the characteristic values as coordinate axes respectively; mapping the first credit and the first characteristic value into a coordinate system to obtain a first coordinate point; determining an early warning level associated with the early warning area of the first coordinate point in the coordinate system as a target early warning level of the user; and performing early warning prompt on the user according to the target early warning level of the user.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to an early warning analysis method, device, equipment, storage medium and product. Background Art

[0002] In the technical fields of big data, artificial intelligence, etc., for operators, predicting users' arrears behavior based on their relevant data is currently a common analysis requirement.

[0003] At present, a common prediction scheme is: based on the user's current remaining balance, monthly package amount, historical average arrears amount and remaining days until the next payment, the monthly subscription package amount, historical average arrears amount and remaining days until the next payment are input into the arrears warning threshold prediction model to obtain the arrears warning threshold, and then based on the current remaining balance and the arrears warning threshold, the user's arrears behavior prediction is triggered. If arrears behavior is predicted, a warning is issued; if no arrears behavior is predicted, no warning is issued.

[0004] However, the current prediction accuracy of user arrears behavior is not high. Therefore, how to provide early warning prompts is a technical problem that needs to be solved urgently. Summary of the invention

[0005] In view of the above problems, the present disclosure is proposed. The present disclosure provides a warning analysis method, device, equipment, storage medium and product.

[0006] According to one aspect of the present disclosure, there is provided a warning analysis method, comprising:

[0007] Performing early warning analysis on the user to obtain a first credit rating and a first characteristic value of the user according to at least one user attribute data of the user, wherein each user attribute data includes attribute data corresponding to at least one user attribute of the user;

[0008] Acquire multiple warning areas in a coordinate system, the multiple warning areas are obtained based on the division of the warning curve in the coordinate system, the multiple warning areas are respectively associated with corresponding warning levels, and the coordinate system is constructed with the credit and the characteristic value as coordinate axes;

[0009] Mapping the first credit and the first characteristic value into the coordinate system to obtain a first coordinate point;

[0010] Determining the warning level associated with the warning area of ​​the first coordinate point in the coordinate system as the target warning level of the user;

[0011] Providing a warning prompt to the user according to the target warning level of the user.

[0012] In addition, according to one aspect of the present disclosure, performing early warning analysis on the user according to at least one user attribute data of the user to obtain the first credit and the first characteristic value of the user includes:

[0013] Performing a credit analysis on the user according to at least one user attribute data of the user to obtain the first credit level of the user;

[0014] Determining target user attribute data of the user according to at least one user attribute data of the user;

[0015] According to the target user attribute data, feature extraction is performed on the user to obtain the first feature value of the user.

[0016] Further, performing credit analysis on the user according to at least one user attribute data of the user to obtain the first credit level of the user includes:

[0017] Calculate the total information entropy of the user according to the attribute data corresponding to at least one user attribute of each user attribute data;

[0018] Calculating the clustering information entropy of the user according to the clustering result of the at least one user attribute data;

[0019] Calculating the difference between the total information entropy and the cluster information entropy to obtain information entropy gain;

[0020] Determining attribute distribution data corresponding to the at least one user attribute;

[0021] The information gain ratio of the user is calculated according to the information entropy gain and the attribute distribution data, and the information gain ratio of the user is determined as the first credit level of the user.

[0022] Furthermore, the calculating of the total information entropy of the user according to the attribute data of each user corresponding to at least one user attribute includes:

[0023] Determine at least one attribute data corresponding to each user attribute according to the attribute data corresponding to at least one user attribute of each user attribute data;

[0024] Calculate the information entropy of each user attribute according to at least one attribute data corresponding to each user attribute;

[0025] The information entropy of each user attribute is weighted and summed to obtain the total information entropy.

[0026] Further, the calculating the information entropy of each user attribute according to at least one attribute data corresponding to each user attribute includes at least one of the following:

[0027] In response to the user attribute being an age attribute, using an age information entropy calculation formula to calculate age information entropy corresponding to at least one age of the age attribute;

[0028] In response to the user attribute being an arrears amount attribute, using an arrears information entropy calculation formula to calculate arrears information entropy corresponding to at least one arrears amount of the arrears amount attribute;

[0029] In response to the user attribute being an arrears duration attribute, using a duration information entropy calculation formula to calculate duration information entropy corresponding to at least one arrears duration of the arrears duration attribute;

[0030] In response to the user attribute being a credit score attribute, using a score information entropy calculation formula to calculate the credit information entropy corresponding to at least one credit score of the credit score attribute;

[0031] In response to the user attribute being an arrears history attribute, a historical information entropy calculation formula is used to calculate a rating information entropy corresponding to at least one historical evaluation level of the arrears history attribute.

[0032] According to one aspect of the present disclosure, the step of calculating the clustering information entropy of the user according to the clustering result of the at least one user attribute data includes:

[0033] Determine a clustering result of the at least one user attribute data, the clustering result comprising clusters corresponding to cluster centers of each user attribute, each cluster containing at least one user attribute data belonging to the same category;

[0034] Calculate the internal information entropy corresponding to at least one user attribute data in each cluster;

[0035] The internal information entropy of each cluster is weighted and summed to obtain the cluster information entropy.

[0036] Furthermore, the calculation formula of the clustering information entropy is as follows:

[0037]

[0038] Among them, H(D) is the clustering information entropy, H(C i ) is the internal information entropy of the i-th user attribute, C i is the number of at least one user attribute data in the i-th cluster, and D is the number of the at least one user attribute data.

[0039] Furthermore, it also includes:

[0040] Initializing a cluster center point of each user attribute according to the at least one user attribute to obtain K cluster center points, where K is the number of the at least one user attribute;

[0041] According to the K cluster centers and a preset clustering algorithm, the at least one user attribute data is clustered to obtain the clustering result, wherein the clustering result includes K clusters, and each cluster includes at least one user attribute data belonging to the same category.

[0042] Further, the determining the attribute distribution data corresponding to the at least one user attribute includes:

[0043] Calculate the attribute distribution data corresponding to the at least one user attribute according to the distribution attribute calculation formula, the distribution attribute calculation formula is as follows:

[0044]

[0045] Among them, Values(A) is all possible values ​​of user attribute A, D v is a subset of data with value v on attribute A, |D v | is the number of samples in the dataset D, |D v | is the data subset D v The number of samples.

[0046] Further, extracting features of the user according to the target user attribute data to obtain the first feature value of the user includes:

[0047] According to the feature extraction formula, the first feature value corresponding to the target user attribute data is calculated. The feature extraction formula is as follows:

[0048]

[0049] Among them, Y is the first characteristic value, and X1-X5 are the attribute data corresponding to at least one user attribute of the target user attribute data.

[0050] According to one aspect of the present disclosure, the invention comprises:

[0051] Constructing the coordinate system with the credit and the characteristic value as coordinate axes respectively;

[0052] Based on a preset warning analysis formula, a warning curve in the coordinate system is determined, wherein the warning curve is used to divide the coordinate system into a plurality of warning areas, each warning area corresponding to a corresponding warning level;

[0053] Mapping the first credit and the first characteristic value into the coordinate system to obtain a first coordinate point includes:

[0054] According to the credit coordinate axis and the characteristic value coordinate axis of the coordinate system, the first coordinate points of the first credit and the first characteristic value in the coordinate system are determined.

[0055] Furthermore, the warning curve in the coordinate system is determined based on the preset warning analysis formula, including:

[0056] Determine a first parameter and a second parameter in the preset early warning analysis formula, wherein the first parameter is a formula in the early warning analysis formula for adjusting the algorithm composition, and the second parameter is an input parameter of the early warning analysis formula;

[0057] Randomly determine a first value of the first parameter and a second value of the second parameter;

[0058] Inputting a first value of the first parameter and a second value of the second parameter into the early warning analysis formula, and calculating and obtaining early warning data corresponding to the first value and the second value;

[0059] Plotting the second parameter and the warning data as coordinate points in the coordinate system to obtain coordinate points in the coordinate system;

[0060] A plurality of coordinate points in the coordinate system are subjected to curve fitting to obtain the warning curve.

[0061] Furthermore, the early warning analysis formula is as follows:

[0062]

[0063] Among them, Y is the second parameter, a, b, c, d, w are all first parameters, and F(Y) is the early warning data obtained by calculation.

[0064] According to one aspect of the present disclosure, the user includes multiple users, and the method further includes:

[0065] Divide users with the same target warning level among multiple users into the same user group to obtain user groups corresponding to each warning level, each user group including at least one user with the same warning level;

[0066] Determine the number of users at each warning level based on the user groups corresponding to each warning level;

[0067] Output the number of users corresponding to each of the plurality of warning levels.

[0068] Furthermore, the outputting of the number of users corresponding to the plurality of warning levels respectively comprises:

[0069] Assign warning colors to the multiple warning levels respectively;

[0070] According to the warning colors corresponding to the warning levels, warning information of the number of users corresponding to the multiple warning levels is generated, and the warning information displays the ratio of the number of at least one user in the corresponding user group to the number of multiple users in the warning color of each warning level.

[0071] According to one aspect of the present disclosure, providing a warning prompt to the user according to the target warning level of the user includes:

[0072] Determining a target prompting mode corresponding to the target warning level according to the prompting modes corresponding to the at least one warning level;

[0073] According to the target prompt mode, generating warning prompt information corresponding to the target warning level;

[0074] Output the warning prompt information.

[0075] As will be described in detail below, according to the early warning analysis method of the embodiment of the present disclosure, a large amount of user attribute data of the user is obtained, and the user attribute data of the user is subjected to early warning analysis to obtain the user's first credit and first characteristic value, so that the first credit can more accurately characterize the user's credit, and the first characteristic value can accurately characterize the user's characteristics. The first credit and the first characteristic value are thus mapped to a coordinate system, and the coordinate system is divided into a plurality of early warning areas. Therefore, the early warning area where the coordinate points of the first credit and the first characteristic value in the coordinate system are located is the final mapping result, and the early warning level associated with the early warning area is the user's target early warning level, and the user is warned through the target early warning level. The user's early warning level can be obtained through the mapping of characteristics and credit, and early warning identification and monitoring in finer dimensions can be achieved through the division of early warning levels, which can effectively improve the accuracy of early warnings.

[0076] It is to be understood that both the foregoing general description and the following detailed description are exemplary, and are intended to provide further explanation of the technology as claimed. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] The above and other purposes, features and advantages of the present disclosure will become more apparent by describing the embodiments of the present disclosure in more detail in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present disclosure and constitute a part of the specification. Together with the embodiments of the present disclosure, they are used to explain the present disclosure and do not constitute a limitation of the present disclosure. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0078] Figure 1 is a schematic diagram illustrating the application of an early warning analysis system according to an embodiment of the present disclosure;

[0079] Figure 2is a flowchart illustrating a warning analysis method according to an embodiment of the present disclosure;

[0080] Figure 3 is an example diagram further illustrating a warning area in the warning analysis method of an embodiment of the present disclosure;

[0081] Figure 4 is a flowchart illustrating another early warning analysis method according to an embodiment of the present disclosure;

[0082] Figure 5 is an example diagram further illustrating the warning prompt information in the warning analysis method of the embodiment of the present disclosure;

[0083] Figure 6 is an example diagram further illustrating an early warning prompt in the early warning analysis method of an embodiment of the present disclosure;

[0084] Figure 7 is a block diagram illustrating an early warning analysis device according to an embodiment of the present disclosure;

[0085] Figure 8 is a hardware block diagram illustrating an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0086] In order to make the purpose, technical solution and advantages of the present disclosure more obvious, the exemplary embodiments according to the present disclosure will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments of the present disclosure, and it should be understood that the present disclosure is not limited to the exemplary embodiments described here.

[0087] The technical solution disclosed in the present invention can be applied to the scenario of predicting user arrears behavior, and more accurate and effective risk assessment can be achieved through the analysis of a large amount of user attribute data and the classification of warning levels.

[0088] In the related art, for operators, user arrears is a common behavior, which usually causes certain economic losses. Predicting the user's arrears behavior and issuing arrears warnings in a timely manner can reduce the phenomenon of arrears and thus reduce risks. At present, the more common arrears prediction method is to use the user's remaining balance, package amount, average arrears amount and other parameters for threshold warnings, that is, when the user's remaining balance is less than a certain threshold, the user is prompted for arrears. However, the accuracy of this arrears prediction is not high.

[0089] In order to solve the above problems, the present disclosure divides the coordinate system into regions to perform early warning classification, and a more detailed dimensional analysis can be achieved through multi-level early warning. In addition, by analyzing the user's arrears behavior, the user's credit and characteristic value are obtained, and then the credit and characteristic value are mapped to the coordinate system to obtain the user's early warning level. A mapping association between credit and characteristic value and early warning level is established to achieve a more accurate mapping of the early warning level. In this solution, a large amount of user-related data participates in the user's early warning analysis, which can achieve effective risk assessment of the user.

[0090] Figure 1 This is an application example diagram of an early warning analysis system provided in an embodiment of the present disclosure. The early warning analysis system may include a user terminal 10 and a server 20. Among them, the user terminal 10 can provide communication services for the user. The server 20 can establish a communication connection with the user terminal 10 and collect user attribute data when the user uses the user terminal 10. The server 20 can be, for example, any type of server such as a cloud server or a background server, and the specific type of the server is not too limited in this embodiment.

[0091] It is understandable that the data of various types or attributes involved in this disclosure are all data authorized by the user and comply with the provisions of relevant laws and regulations.

[0092] The server 20 may be configured with the early warning analysis method of the present disclosure and execute the early warning analysis method, so as to provide an early warning prompt to the user after obtaining the user's target early warning level.

[0093] Figure 2 This is a flowchart of a warning analysis method provided by an embodiment of the present disclosure. The warning analysis method can be configured as a warning analysis device, which can be located in an electronic device (such as a cloud server). The warning analysis method can include the following steps:

[0094] S201. Perform early warning analysis on the user to obtain a first credit rating and a first characteristic value of the user based on at least one user attribute data of the user, wherein each user attribute data includes attribute data corresponding to at least one user attribute of the user.

[0095] As an embodiment, before S201, the method further includes: collecting at least one user attribute data of the user. Specifically, collecting at least one user attribute data of the user in a target period. The target period may refer to a predetermined time period, and the start time and end time of the target period may be obtained by setting.

[0096] For example, a time period setting page may be displayed, the time period setting page may include setting controls corresponding to a start time and an end time respectively, in response to an input operation on the setting control for the start time, the start time is obtained, and in response to an input operation on the setting control for the end time, the end time is obtained. The target time period corresponding to the start time and the end time is determined.

[0097] It is understandable that the start time and end time of the target period can be set according to the needs. For example, at least one user attribute data of the user within 3 years can be collected. Taking the end time as the current time as an example, the start time can be the current time + 3 years.

[0098] Optionally, collecting at least one user attribute data of the user may include: collecting at least one user attribute data of the user according to a preset target period and a collection frequency. For example, the target period may be 3 years, and the collection frequency may be once a month. Therefore, the user attribute data of the user may be collected every month within 3 years, i.e., 12*3=36 user attribute data may be included.

[0099] Optionally, the user attribute data may refer to user personalized attribute data, and may include attribute data corresponding to at least one user attribute. The user attribute may refer to an attribute for detecting user behavior, such as at least one of the following user attributes: age attribute, arrears amount attribute, arrears duration attribute, credit score attribute, or arrears history attribute. The attribute data corresponding to at least one user attribute may include, for example, the arrears amount corresponding to the arrears amount attribute, the age corresponding to the age attribute, the arrears duration corresponding to the arrears duration attribute, the credit score corresponding to the credit score attribute, and the historical evaluation level corresponding to the arrears history attribute.

[0100] Optionally, the first credit rating may refer to a value used by the user in the operator's operating system to characterize the credit rating, such as -1, 0, 1, etc. The credit rating range and specific value may be set according to usage requirements, such as a ten-point system or a hundred-point system.

[0101] Optionally, the first eigenvalue may refer to feature data obtained by analyzing the user's behavior or characteristics according to the attribute data of each user attribute. For example, the eigenvalue may be an eigenvalue, an eigenvector or a eigenmatrix.

[0102] Optionally, S201 may include: performing credit analysis on the user according to at least one user attribute data of the user to obtain a first credit rating of the user. Performing feature extraction on the user according to at least one user attribute data of the user to obtain a first feature value of the user.

[0103] S202, obtaining a plurality of warning areas in a coordinate system, wherein the plurality of warning areas are obtained based on the division of warning curves in the coordinate system, the plurality of warning areas are respectively associated with corresponding warning levels, and the coordinate system is constructed with the credit and the characteristic value as coordinate axes.

[0104] Optionally, multiple warning areas can be pre-set in the coordinate system, such as performing a region division operation on the coordinate system according to the warning curve in the coordinate system to obtain multiple warning areas. The region division operation may refer to a division operation performed by the user on the overall area of ​​the coordinate system according to the warning curve, such as a line segment drawing operation performed on multiple division points in the warning curve of the coordinate system.

[0105] Specifically, performing a region division operation on the coordinate system according to the warning curve in the coordinate system to obtain multiple warning regions may include: selecting multiple division points according to the warning curve in the coordinate system, performing a division operation on the region of the coordinate system according to the multiple division points, and obtaining multiple warning regions.

[0106] The region of the coordinate system may refer to, for example, a region formed by a value range corresponding to the credit level and a value range corresponding to the characteristic value.

[0107] Furthermore, performing a division operation on an area of ​​the coordinate system according to multiple division points to obtain multiple warning areas may include: drawing straight lines parallel to the credit coordinate axis and passing through each division point, and obtaining multiple warning areas obtained by dividing the area of ​​the coordinate system with multiple straight lines parallel to the credit coordinate axis.

[0108] Specifically, the warning curve may be an S-shaped curve, and selecting a plurality of division points according to the warning curve in the coordinate system may include: using a plurality of inflection points in the S-shaped warning curve as division points respectively.

[0109] For ease of understanding, Figure 3 An example diagram of early warning area division is shown. Figure 3 As shown, the warning curve is 301, and there are multiple dividing points 302 on the warning curve. Figure 3 The multiple coordinate points in the S-shaped warning curve where the curve slope change is greater than the preset slope threshold value can be used as division points 302. Then, a straight line 303 passing through each division point 302 and parallel to the credit coordinate axis is drawn, and the multiple straight lines 303 divide the area of ​​the coordinate system into multiple warning areas.

[0110] Optionally, the coordinate system may be a two-dimensional coordinate system, and the two coordinate axes of the two-dimensional coordinate system may respectively refer to the credit and the characteristic value, the coordinate axis referring to the credit may be called the credit coordinate axis, and the coordinate axis referring to the characteristic value may be called the characteristic value coordinate axis. For example, the credit coordinate axis may be the Y axis, and the characteristic value coordinate axis may be the X axis.

[0111] S203: Map the first credit and the first characteristic value into a coordinate system to obtain a first coordinate point.

[0112] Optionally, S203 may include: determining coordinate points of the first credit and the first characteristic value in the coordinate system to obtain the first coordinate point. That is, the value corresponding to the first coordinate point in the credit coordinate system is the first credit, and the value corresponding to the first coordinate point in the characteristic value coordinate system is the first characteristic value.

[0113] S204: Determine the warning level associated with the warning area of ​​the first coordinate point in the coordinate system as the target warning level of the user.

[0114] Optionally, the warning area of ​​the first coordinate point in the coordinate system can be determined in the following manner: obtain the value range of each warning area in the coordinate system, determine the target value range that includes the first coordinate point based on the value range of each warning area, and determine the warning area corresponding to the target value range as the warning area of ​​the first coordinate point in the coordinate system.

[0115] S205: Prompt a warning to the user according to the user's target warning level.

[0116] Optionally, S205 may include: generating target warning information of the user according to the target warning level of the user, and outputting the target warning information of the user.

[0117] Further, according to the target warning level of the user, generating the target warning information of the user may include: searching for a target warning template that matches the target warning level of the user, filling the user's arrears information into the target warning template, and obtaining the target warning information. The user's arrears information may include, for example, at least one item such as the amount of arrears, the content of arrears, the duration of arrears, arrears retention, or product recommendation information.

[0118] Among them, the warning template may refer to a pre-set template for generating warning information, such as a text template, an email template, a collection letter, a text message template, a prompt page of an application, etc.

[0119] Optionally, the warning prompts for users at different warning levels can be different. Assuming that the warning levels include levels one to five, personalized warning prompts and product recommendations can be set for users at each warning level. From the perspective of arrears warning analysis, this article explains in more detail how to conduct arrears analysis and strategy configuration, achieve accurate arrears retention and marketing work, and improve customer retention and arrears recovery efficiency. Specific recommendation methods include but are not limited to the following:

[0120] Level 1 users: Targeted warning information includes relief for overdue bills and promotional activities to enhance customer satisfaction and increase customer retention.

[0121] Secondary users: Targeted warning information includes customized arrears installment and repayment plans based on the amount of arrears and arrears history to promote arrears repayment behavior.

[0122] Level 3 users: Targeted warning information includes overdue bill details and overdue bill reminders to guide customers to pay overdue bills in a timely manner.

[0123] Level 4 users: Targeted warning information is manual prompts, such as reminders of overdue payments via text messages and phone calls, to encourage customers to actively deal with overdue payment issues.

[0124] Level 5 users: Targeted warning information is manual prompts, such as personalized customer return visits and consulting services to reactivate overdue customers and reduce the overdue churn rate.

[0125] In the technical solution disclosed in the present invention, a large amount of user attribute data of the user is obtained, and the user attribute data of the user is subjected to early warning analysis to obtain the user's first credit and first characteristic value, so that the first credit can more accurately characterize the user's credit, and the first characteristic value can accurately characterize the user's characteristics. The first credit and the first characteristic value are thus mapped to a coordinate system, and the coordinate system is divided into a plurality of early warning areas. Therefore, the early warning area where the coordinate points of the first credit and the first characteristic value in the coordinate system are located is the final mapping result, and the early warning level associated with the early warning area is the user's target early warning level, and the user is warned through the target early warning level. The user's early warning level can be obtained through the mapping of characteristics and credit, and early warning identification and monitoring in finer dimensions can be achieved through the division of early warning levels, which can effectively improve the accuracy of early warnings.

[0126] Further, based on any of the above embodiments, according to at least one user attribute data of the user, a warning analysis is performed on the user to obtain a first credit rating and a first characteristic value of the user, including:

[0127] Performing a credit analysis on the user according to at least one user attribute data of the user to obtain a first credit rating of the user;

[0128] Determining target user attribute data of the user according to at least one user attribute data of the user;

[0129] According to the target user attribute data, the user's features are extracted to obtain the user's first feature value.

[0130] Optionally, determining the target user attribute data of the user based on at least one user attribute data of the user may include: determining the user attribute data collected last time in at least one user attribute data of the user as the target user attribute data. Alternatively, determining any one user attribute data in at least one user attribute data of the user as the target user attribute data. Alternatively, calculating the average value of each user attribute based on the attribute data corresponding to at least one user attribute of each user attribute data, and determining the average value of each user attribute as the target user attribute data. For example, the average amount of arrears corresponding to at least one amount of arrears may be calculated, the average duration of arrears corresponding to at least one duration of arrears may be calculated, the average score corresponding to at least one credit score may be calculated, the average rating corresponding to at least one arrears history information may be calculated, and the average age corresponding to at least one age may be calculated.

[0131] Among them, for the arrears history information represented by the arrears level, each arrears level can be converted into a rating value, such as good is converted to 3, medium is converted to 2, and poor is converted to 1. Then the rating values ​​corresponding to each arrears history information are averaged to obtain the average rating.

[0132] Of course, when calculating the average of the duration of arrears, score or age, you can just add the sum and then divide by the number of data.

[0133] Optionally, performing credit analysis on the user according to at least one user attribute data of the user to obtain the first credit of the user may include: extracting information entropy of at least one user attribute data to obtain information entropy gain, clustering at least one user attribute data to obtain target cluster data, and calculating the quotient of the information entropy gain and the target cluster data to obtain the first credit.

[0134] Optionally, extracting the information entropy of at least one user attribute data to obtain the information entropy gain may include calculating the total information entropy of the user according to the attribute data corresponding to at least one user attribute of each user attribute data; calculating the cluster information entropy of the user according to the total information entropy; and calculating the difference between the total information entropy and the cluster information entropy to obtain the information entropy gain.

[0135] In the technical solution disclosed in the present invention, when performing early warning analysis on a user, a credit analysis can be performed on the user based on at least one user attribute data to obtain a first credit rating, thereby realizing a credit analysis based on the user attribute data and obtaining an accurate first credit rating. As for feature extraction, target user attribute data is determined based on at least one user attribute data, and the target user attribute data is used to extract the user's features, thereby obtaining the user's first feature value and improving the user's feature extraction efficiency.

[0136] Further, based on any of the above embodiments, performing a credit analysis on the user according to at least one user attribute data of the user to obtain a first credit level of the user includes:

[0137] Calculate the total information entropy of the user according to the attribute data corresponding to at least one user attribute of each user attribute data;

[0138] Calculating the user's clustering information entropy based on the clustering result of at least one user attribute data;

[0139] Calculate the difference between the total information entropy and the clustering information entropy to obtain the information entropy gain;

[0140] Determining attribute distribution data corresponding to at least one user attribute data;

[0141] The quotient of information entropy gain and attribute distribution data is calculated to obtain the user's first credit.

[0142] It is understandable that the total information entropy may refer to the sum of the information entropies of each user attribute, such as being obtained by weighted summation of the information entropies of each user attribute.

[0143] Furthermore, calculating the total information entropy of the user based on the attribute data corresponding to at least one user attribute of each user attribute data may include: calculating the information entropy of each user attribute based on the attribute data corresponding to at least one user attribute of each user attribute data, and weighted summing the information entropies of each user attribute to obtain the total information entropy.

[0144] Optionally, calculating the user's clustering information entropy based on the clustering result of at least one user attribute data may include: determining the internal information entropy of each category based on the clustering result of at least one user attribute data, and weighted summing the internal information entropies of each category to obtain the user's clustering information entropy.

[0145] Optionally, calculating the difference between the total information entropy and the cluster information entropy to obtain the information entropy gain may include: calculating the difference between the total information entropy and the cluster information entropy, and determining the absolute value of the difference as the information entropy gain.

[0146] It can be understood that the attribute distribution data of at least one user attribute data is inherent data of the distribution of at least one user attribute, and can be obtained by calculation using a distribution attribute calculation formula.

[0147] In the technical solution disclosed in the present invention, the total information entropy of the attribute data of each user attribute can be extracted first, and the total information entropy can characterize the amount of information contained in at least one user attribute data as a whole. After that, the clustering information entropy of at least one user attribute data is calculated, and the data with the same user attribute is classified by clustering, and then the clustering information entropy is calculated using the clustering result. The information entropy calculation of at least one user attribute data is completed in two ways, and then the total information entropy and clustering information entropy are used to calculate the information gain, which can reflect the credit risk of the user in each user attribute and realize a more comprehensive information entropy analysis. Therefore, after obtaining the information entropy gain, a more comprehensive and comprehensive credit system evaluation is formed, and a more accurate credit evaluation result is obtained.

[0148] In a possible design, the total information entropy of the user is calculated according to the attribute data corresponding to at least one user attribute of each user attribute data, including:

[0149] Determine at least one attribute data corresponding to each user attribute according to the attribute data corresponding to at least one user attribute of each user attribute data;

[0150] Calculate the information entropy of each user attribute according to at least one attribute data corresponding to each user attribute;

[0151] The information entropy of each user attribute is weighted and summed to obtain the total information entropy.

[0152] Optionally, the attribute data of each user attribute data respectively corresponding to at least one user attribute may be determined as at least one attribute data corresponding to the user attribute. That is, the data belonging to each user attribute in each user attribute data is directly obtained.

[0153] Furthermore, after obtaining at least one attribute data of each user attribute, at least one attribute data of each user attribute may be calculated to obtain information entropy of each user attribute.

[0154] Optionally, an information entropy calculation formula may be used to calculate the information entropy of at least one attribute data of each user attribute. The information entropy calculation formula may be referred to in the following content and will not be described in detail here.

[0155] In the technical solution disclosed in the present invention, when calculating the total information entropy of a user, all attribute data related to the user's attributes are used to calculate the information entropy, so that more effective information entropy can be used to accurately evaluate the credit impact of the corresponding user attributes and obtain a more comprehensive evaluation result.

[0156] Further, on the basis of any of the above embodiments, the information entropy of each user attribute is calculated according to at least one attribute data corresponding to each user attribute, including at least one of the following:

[0157] In response to the user attribute being an age attribute, using an age information entropy calculation formula to calculate the age information entropy corresponding to at least one age of the age attribute;

[0158] In response to the user attribute being an arrears amount attribute, using an arrears information entropy calculation formula to calculate arrears information entropy corresponding to at least one arrears amount of the arrears amount attribute;

[0159] In response to the user attribute being an overdue duration attribute, using a duration information entropy calculation formula to calculate duration information entropy corresponding to at least one overdue duration of the overdue duration attribute;

[0160] In response to the user attribute being a credit score attribute, using a score information entropy calculation formula to calculate the credit information entropy corresponding to at least one credit score of the credit score attribute;

[0161] In response to the user attribute being an arrears history attribute, a rating information entropy corresponding to at least one historical evaluation level of the arrears history attribute is calculated using a historical information entropy calculation formula.

[0162] Optionally, the age information entropy calculation formula is as follows:

[0163]

[0164] Among them, P age，k is the probability of the kth case of the age attribute, n age is the possible value of the age attribute. For example, if at least one of the ages corresponding to the age attribute in the collected user attribute data is 10, 20, or 30, then n age Includes the above 3 age attributes.

[0165] It is understandable that the age information entropy calculation formula introduces logarithmic function, square root, power operation and constant term addition. These operations combined increase the complexity of the calculation and better reflect the influence of age attributes on the clustering results. The logarithmic function represents the logarithm with base 3, and the square is the square root of 2. These operations increase the specificity and innovation of the formula.

[0166] In addition, age may be discrete, and there may be specific relationships between age groups. Adding logarithmic functions and squares as weights can better distinguish the impact of age groups and improve the sensitivity of clustering results.

[0167] Optionally, the calculation formula for the arrears information entropy is as follows:

[0168]

[0169] Among them, P amountowed,k is the probability of the kth case of the arrears amount attribute, n amountowedis the possible number of the arrears amount attribute. For example, the arrears amount attribute can be segmented or histogrammed. amountowed It can be the number of segments of the arrears amount or the number of different arrears amounts counted in the histogram.

[0170] It is understandable that the calculation formula for arrears information entropy introduces logarithmic functions, exponential functions, power operations, and the addition of constant terms. These operations combined increase the complexity of the calculation and better reflect the degree of influence of the arrears amount attribute on the clustering results. The logarithmic function log2 represents the logarithm with base 2, and the exponential function ek represents the exponent with base e as the natural constant. These operations increase the uniqueness and innovation of the formula.

[0171] In addition, the amount of outstanding payments may present a long-tail distribution, and a larger amount may have a greater impact on the clustering results. Adding logarithmic functions and cubic functions as weights can better distinguish the impact of the amount and highlight the importance of large outstanding payments.

[0172] Optionally, the duration information entropy calculation formula is as follows:

[0173]

[0174] Among them, P overdueduration,k is the probability of the kth case of the arrears duration attribute, n overdueduraion is the number of possible values ​​of the arrears duration attribute. For example, the arrears duration can be segmented or histogrammed. overdueduration It can be the number of segments of the arrears duration, or the number of different arrears durations obtained by histogram statistics.

[0175] It is understandable that the calculation formula of duration information entropy introduces square root, logarithmic function, exponential function and power operation. These operations combined increase the complexity of calculation and better reflect the influence of the arrears duration attribute on the clustering results. Square root and logarithmic function make the influence on the long duration more significant, and exponential function and power operation enhance the distinction of the influence of different durations.

[0176] In addition, the duration of arrears may be related to the amount of arrears and the history of arrears, and a longer duration may more seriously affect the user's credit status. Adding the reciprocal and square as weights can better reflect the impact of the duration on the clustering results and increase the sensitivity to long-term arrears.

[0177] Optionally, the scoring information entropy calculation formula is as follows:

[0178]

[0179] Among them, P creditscore,k is the probability of the kth case of the credit score attribute, such as the probability of a credit score of 750.creditscore is the number of possible values ​​of the credit score attribute. If the credit score can be segmented or histogram statistics, n creditscore This can be the number of segments, or a histogram of the number of different scores obtained.

[0180] It is understandable that the scoring information entropy formula introduces logarithmic functions, square roots, exponential functions, and power operations. These operations combined increase the complexity of the calculation and better reflect the degree of influence of credit score attributes on clustering results. The logarithmic function represents the logarithm with base 2, and the exponential function represents the exponent with base e as the natural constant.

[0181] In addition, credit scores may be continuous, and high and low scores may have different degrees of impact on clustering results. Adding exponential functions as weights can better reflect the impact of scores and increase the sensitivity of credit scores.

[0182] Optionally, the historical information entropy calculation formula is as follows:

[0183]

[0184] Among them, P paymenthistory，k is the probability of the kth case of the arrears history attribute, n paymenthistory is the number of possible values ​​of the arrears history attribute. For example, if the arrears history includes good, medium, and poor, n paymenthistory is 3.

[0185] Optionally, the historical information entropy calculation formula introduces logarithmic functions, exponential functions, power operations, and addition of constant terms. These operations combined increase the complexity of the calculation and better reflect the degree of influence of the arrears history attribute on the clustering result. The power operation represents the square root of k, and the exponential function represents the negative kth power of e.

[0186] In addition, the history of arrears may have a significant impact on the user's credit status, and a long-term history of arrears may be more serious. Adding square roots and cubes as weights can better reflect the degree of historical impact and increase the sensitivity to a history of arrears.

[0187] Optionally, weighted summing the information entropies of each user attribute to obtain the total information entropy may include: weighted summing the age information entropy, arrears information entropy, duration information entropy, credit information entropy and rating information entropy to obtain the total information entropy.

[0188] Exemplarily, the weighted formula of the total information entropy is as follows:

[0189] H(C)=ω1*H age +ω2*H amountowed +ω3*H overdueduration +ω4*H creditscore+ω5*H paymenthistory

[0190] Among them, ω1 is the weight of the age attribute, ω2 is the weight of the arrears amount attribute, ω3 is the weight of the arrears duration attribute, ω4 is the weight of the credit score attribute, and ω5 is the weight of the arrears history attribute.

[0191] In the technical solution disclosed in the present invention, at least one of the age attribute, the arrears amount attribute, the arrears duration attribute, the credit score attribute and the arrears history attribute is used to evaluate the user's credit. Through more detailed division of fields and clearer credit evaluation, it is possible to evaluate the user's credit from multiple dimensions.

[0192] As an embodiment, calculating the clustering information entropy of the user according to the clustering result of at least one user attribute data includes:

[0193] Determine a clustering result of at least one user attribute data, the clustering result includes clusters corresponding to cluster centers of each user attribute, each cluster contains at least one user attribute data belonging to the same category;

[0194] Calculate the internal information entropy corresponding to at least one user attribute data in each cluster;

[0195] The internal information entropy of each cluster is weighted and summed to obtain the cluster information entropy.

[0196] Optionally, determining the clustering result of at least one user attribute data may include: obtaining a clustering result obtained by clustering the at least one user attribute data. Specifically, the clustering result may be obtained by clustering the at least one user attribute data using a clustering algorithm.

[0197] Optionally, the clustering result includes K clusters, where K is the number of attributes of the at least one user attribute.

[0198] In the technical solution of the present invention, at least one user attribute data can be clustered, and the obtained clustering result can protect the cluster corresponding to the cluster center of each user attribute, and the cluster contains at least one user attribute belonging to the same category. Using the clustering method to cluster at least one user attribute data can obtain at least one user attribute data with the same or similar user attributes, and then calculate the internal information entropy of at least one user attribute data in each cluster. The internal information entropy can more accurately characterize the credit characteristics of the user in the corresponding user attribute, thereby ensuring that the cluster information entropy obtained by weighted summation of the internal information entropy of each cluster can reflect the impact of each user attribute on the user credit, thereby improving the characterization effect of the cluster information entropy on the user credit.

[0199] In one possible design, the clustering information entropy calculation formula is as follows:

[0200]

[0201] Among them, H(D) is the clustering information entropy, H(C i ) is the internal information entropy of the i-th user attribute, C i is the number of at least one user attribute data in the i-th cluster, and D is the number of at least one user attribute data.

[0202] As an embodiment, in order to implement clustering of at least one user attribute data, the following steps are also included:

[0203] Initialize the cluster center point of each user attribute according to at least one user attribute to obtain K cluster center points, where K is the number of at least one user attribute;

[0204] At least one user attribute data is clustered according to K cluster centers and a preset clustering algorithm to obtain a clustering result, where the clustering result includes K clusters, and each cluster includes at least one user attribute data belonging to the same category.

[0205] Optionally, initializing the cluster center of each user attribute according to at least one user attribute to obtain K cluster centers may include: determining the number of clusters K according to at least one user attribute, and initializing the K cluster centers. The number of clusters K may be the number of at least one user attribute.

[0206] Optionally, the method may further include: inputting at least one user attribute data into a clustering algorithm to obtain clustering information output by the clustering algorithm. The clustering information may include clusters of the K clustering centers.

[0207] Optionally, the clustering algorithm may be a centered clustering algorithm or a decentered clustering algorithm, and specifically may be a K-means clustering algorithm, etc. The specific type of the clustering algorithm is not excessively limited in this embodiment.

[0208] For example, a clustering algorithm may include the following steps:

[0209] S1. Randomly select K initial cluster centers, or use the startup method to select K initial cluster centers.

[0210] S2. Determine a data set D consisting of at least one user attribute data.

[0211] S3. Calculate the distances between each user attribute data in the data set D and the K initial cluster centers, and assign each user attribute data to the cluster center with the smallest distance, to obtain at least one user attribute data assigned to each cluster center.

[0212] S4. Taking at least one user attribute data assigned to each cluster center as a cluster, calculating the centroid of each cluster, and updating the centroid of each cluster as a new cluster center, and returning to S3 to continue execution until a preset convergence condition is met.

[0213] Optionally, after the clustering algorithm is finished, clustering information can be determined, and the clustering information can include the final N clusters and at least one user attribute data in each cluster. After obtaining the clustering information, the internal information entropy of each cluster can also be calculated to obtain the information entropy of each cluster.

[0214] In a possible design, the number of attributes of at least one user attribute can be used as the number of initial cluster centers, that is, K can be taken as the number of attributes of at least one user attribute. Then, clustering is performed with each user attribute as the center.

[0215] For example, assuming there are five user attributes, namely, age attribute, arrears amount attribute, arrears duration attribute, credit score attribute, and arrears history attribute, five initial clustering centers can be set first.

[0216] For each cluster center, a cluster center matching the corresponding user attribute is set, thereby obtaining a cluster of each user attribute, the result of which is at least one user attribute data with the same user attribute.

[0217] After obtaining the cluster of each user attribute, information entropy calculation may be performed on at least one user attribute data in each user attribute cluster to obtain internal information entropy of each user attribute.

[0218] It is understandable that after clustering, the information entropy of at least one user attribute data in each user attribute cluster may be calculated using an existing information entropy calculation formula, and this is not particularly limited in this embodiment.

[0219] In the technical solution disclosed in the present invention, when clustering at least one user attribute data, the user attribute can be used as the clustering center for clustering, that is, at least one user attribute involved in user data collection is involved in the clustering process, and the obtained clustering result can classify the user attribute data from the dimensions corresponding to each user attribute, reflecting the uncertainty of different user attributes in expressing user credit, thereby further improving the prediction effect of user credit.

[0220] In one possible design, determining attribute distribution data corresponding to at least one user attribute data includes:

[0221] According to the distribution attribute calculation formula, attribute distribution data corresponding to at least one user attribute data is determined. The distribution attribute calculation formula is as follows:

[0222]

[0223] Among them, Values(A) is all possible values ​​of user attribute A, D v is the data subset with value v on attribute A, |D| is the number of samples in data set D, |D v | is the data subset D v The number of samples.

[0224] Furthermore, Values(A) may refer to user attributes A, such as age attribute, overdue amount attribute, overdue duration attribute, credit score attribute, and overdue history attribute.

[0225] Optionally, when calculating the attribute distribution data corresponding to at least one user attribute data by using the distributed attribute calculation formula, a clustering result of at least one user attribute data may be obtained first, the clustering result includes clusters corresponding to each user attribute, each cluster includes at least one user attribute data having the same user attribute. The user attribute data with a value of V in the cluster of each user attribute A is determined as the data subset of the user attribute A.

[0226] The number of samples can refer to the number of data in a data set. For example, if a data set contains 100 samples, the number of samples is 100.

[0227] In the technical solution disclosed in the present invention, a distributed attribute calculation formula is adopted to accurately calculate the attribute distribution data of the user.

[0228] As an embodiment, extracting features of a user according to target user attribute data to obtain a first feature value of the user includes:

[0229] According to the feature extraction formula, the first feature value corresponding to the target user attribute data is calculated. The feature extraction formula is as follows:

[0230]

[0231] Among them, Y is the first characteristic value, and X1-X5 are attribute data corresponding to at least one user attribute of the target user attribute data.

[0232] For example, X1 can be age, X2 can be the amount of arrears, X3 can be the duration of arrears, X4 can be the credit score, and X5 can be the history of arrears. Through product terms such as X1-X5, the interaction between features is introduced, and the impact of the joint action of features on risk assessment is considered.

[0233] Optionally, calculating the first feature value corresponding to the target user attribute data according to the feature extraction formula may include inputting the target user attribute data into the feature extraction formula to obtain the first feature value.

[0234] In the technical solution disclosed in the present invention, a preset feature extraction formula is used to extract the first eigenvalue in the target user attribute data, so as to extract the user's features from multiple user attributes, so that the user's feature expression dimension is richer, and the obtained first eigenvalue can more accurately characterize the user's characteristics.

[0235] As an embodiment, in order to establish a coordinate system, the following steps are also included:

[0236] Construct a coordinate system with credit and eigenvalue as coordinate axes respectively;

[0237] Based on the preset warning analysis formula, a warning curve in the coordinate system is determined. The warning curve is used to divide the coordinate system into multiple warning areas, and each warning area corresponds to a corresponding warning level;

[0238] Mapping the first credit and the first characteristic value into a coordinate system to obtain a first coordinate point includes:

[0239] According to the credit coordinate axis and the eigenvalue coordinate axis of the coordinate system, the first coordinate point of the first credit and the first eigenvalue in the coordinate system is determined.

[0240] Optionally, in the process of dividing the coordinate system into multiple warning areas through the warning curve, the division can be performed according to the value range of the credit level. For example, assuming that the value of Y is 1 and the credit level is represented by P, the multiple warning levels can be divided as follows:

[0241] Level 1: P(Y=1)<0.4. Level 1 means the risk is low, everything is normal and no special precautions are needed.

[0242] Level 2: 0.4 <= P(Y=1) < 0.8, indicating low to moderate risk, with some potential problems or trends. It is recommended to strengthen monitoring and pay attention, and some preventive measures may need to be taken.

[0243] Level 3: 0.8<=P(Y=1)<1.2, indicating a high risk, with obvious problems or trends. Urgent measures need to be taken to prevent the problem from getting worse.

[0244] Level 4: 1.2 <= P(Y=1) < 1.6, indicating that the risk is very high and a serious problem or crisis may have occurred. Urgent and strong actions need to be taken immediately to prevent further losses.

[0245] Level 5: 1.6<P(Y=1)> =2, indicating extreme risks or disasters that may have caused irreversible serious consequences and require immediate and stringent measures for emergency rescue and response.

[0246] In the technical solution disclosed in the present invention, when constructing a coordinate system, the coordinate system is constructed with the credit and the eigenvalue as the coordinate axes, so that the coordinate system can perform data analysis from two dimensions: the credit and the eigenvalue. The warning curve can represent the changes in the user's credit and the eigenvalue. Therefore, using the warning curve to divide the warning area can obtain multiple warning areas with a high division effect, and each warning area is associated with a warning level. Therefore, after the first coordinate point is obtained by mapping the first credit and the first eigenvalue through the coordinate system, the user's warning level can be determined by the warning area where the first coordinate point is located. This makes the process of obtaining the user's warning level more efficient and the target warning level obtained more accurate.

[0247] In a possible design, based on a preset early warning analysis formula, the early warning curve in the coordinate system is determined to include:

[0248] Determine a first parameter and a second parameter in a preset early warning analysis formula, wherein the first parameter is a formula used to adjust the algorithm composition in the early warning analysis formula, and the second parameter is an input parameter of the early warning analysis formula;

[0249] Randomly determine a first value of the first parameter and a second value of the second parameter;

[0250] Inputting a first value of the first parameter and a second value of the second parameter into a warning analysis formula, and calculating to obtain warning data corresponding to the first value and the second value;

[0251] The second value of the second parameter and the warning data are plotted as coordinate points in the coordinate system to obtain the coordinate points in the coordinate system;

[0252] Perform curve fitting on multiple coordinate points in the coordinate system to obtain a warning curve.

[0253] In the disclosed embodiment, when drawing the warning curve, the first parameter and the second parameter in the warning analysis formula can be obtained first, and the first parameter and the second parameter can be set to obtain the first value and the second value, so as to input the obtained first value and the second value into the warning analysis formula to obtain the corresponding warning data. That is, the curve is drawn by taking and inputting data. Thereby, the second value of the second parameter and the warning data are drawn as coordinate points into the coordinate system to obtain the coordinate points in the coordinate system, and the multiple coordinate points in the coordinate system can be curve fitted to obtain the warning curve. The warning curve can clearly show the relationship between the user's credit and the characteristics, so that the warning area division performed by the warning curve is more accurate and has a more accurate zoning effect.

[0254] As an optional implementation method, the early warning analysis formula is as follows:

[0255]

[0256] Among them, Y is the second parameter, a, b, c, d, w are all first parameters, and F(Y) is the early warning data obtained by calculation.

[0257] For example, Y represents the characteristic value or trigger condition of the arrears signal, and the value range of Y can be -10<=Y<=10.

[0258] When the value of Y is known, the value range of a is 0.5 to 2, which is adjusted according to the sensitivity of the arrears signal. A larger value indicates a faster response to the arrears signal. The value range of b is -10 to 0, which is adjusted according to the threshold of the arrears signal. A smaller value indicates that the arrears warning is issued earlier. The value range of C is 0.5 to 2, which is adjusted according to the slope of the arrears signal. A larger value indicates that the arrears signal changes more steeply. The recommended value range of D is -10 to 0, which is adjusted according to the threshold position of the arrears signal. A smaller value indicates that the arrears warning is issued earlier. The value range of w is 0.5 to 2, which is adjusted according to the urgency of the arrears warning. A larger value indicates that the urgency of the arrears warning is higher.

[0259] In the technical solution disclosed in the present invention, the early warning analysis formula can use two parameters. The early warning analysis formula uses two sets of parameters for adjustment, comprehensively considers the impact of multiple variables on the early warning results, and improves the accuracy of the early warning. The exponential function part in the formula enables the early warning results to achieve smooth transition in different intervals. This avoids sudden changes in the early warning results and improves the reliability of the early warning.

[0260] In a possible design, the user includes multiple users, and the method further includes:

[0261] Clustering the target warning levels corresponding to the multiple users respectively to obtain user groups corresponding to the multiple warning levels respectively, wherein the user group includes at least one user with the same warning level;

[0262] Determine the number of users at each warning level based on the user groups corresponding to each warning level;

[0263] Output the number of users corresponding to multiple warning levels.

[0264] In the technical solution disclosed in the present invention, by analyzing the behavior of users at different warning levels, the credit risk of users can be better understood. Thus, the warning levels of users are graded to better identify potential risks. Users at high warning levels may face greater risks, and timely reminders can prompt them to take preventive measures to reduce the possibility of risk occurrence. The timeliness and effectiveness of risk warning reminders are improved.

[0265] like Figure 4 As shown, it is another flow chart of a warning analysis method provided by an embodiment of the present disclosure, and the warning analysis method may include the following steps:

[0266] S401, collecting a plurality of user attribute data corresponding to a plurality of users, each user attribute data including attribute data corresponding to at least one user attribute;

[0267] S402: Performing behavior warning analysis and processing on each user according to the user attribute data of each user to obtain a first credit rating and a first characteristic value of each user;

[0268] S403, determining a coordinate system constructed with the credit and the characteristic value as coordinate axes;

[0269] S404. Based on a preset warning analysis formula, determine a warning curve in the coordinate system. The warning curve is used to divide the coordinate system into multiple warning areas, and each warning area corresponds to a corresponding warning level.

[0270] S405: Determine the first coordinate point of the first credit and the first characteristic value of each user in the coordinate system.

[0271] S406. Determine the target warning level of each user according to the warning level associated with the warning area of ​​the first coordinate point of each user in the coordinate system.

[0272] S407: users with the same target warning level among the multiple users are divided into the same user group to obtain user groups corresponding to each warning level, each user group including at least one user with the same warning level.

[0273] S408. Determine the number of users at each warning level according to the user groups corresponding to each warning level.

[0274] S409: Output the number of users corresponding to the multiple warning levels.

[0275] In the technical solution disclosed in the present invention, when there are multiple users, early warning analysis can be performed on the multiple users respectively to obtain the target early warning levels corresponding to the multiple users respectively, and then the users with the same target early warning level can be divided into the same user group to achieve the classification of users at the same level, so as to provide targeted prompts to users at the same early warning level, such as prompts for the number of users at each early warning level. Timely prompts for the number of users at each early warning level can enable the backend users to make corresponding processing in time, avoiding losses caused by untimely prompts or processing.

[0276] As an embodiment, outputting the number of users corresponding to multiple warning levels respectively includes:

[0277] Assign warning colors to multiple warning levels respectively;

[0278] According to the warning colors corresponding to each warning level, warning information of the number of users corresponding to multiple warning levels is generated, and the warning information displays the ratio of the number of at least one user in the corresponding user group to the number of multiple users in the warning color of each warning level.

[0279] Optionally, respectively allocating warning colors to a plurality of warning levels may, for example, refer to determining color values ​​corresponding to the plurality of warning levels, where the color values ​​are, for example, RGB values.

[0280] For example, assuming there are 5 warning levels, the first level can be set to blue warning, the second level can be set to yellow warning, the third level can be set to orange warning, the fourth level can be set to red warning, and the fifth level can be set to purple warning.

[0281] Optionally, the warning information may be a warning data table, a warning graph, or a warning page. Providing warnings through data analysis of user behavior can effectively reduce operational risks.

[0282] Exemplarily, warning prompts are given to the number of users corresponding to the multiple warning levels according to the warning colors corresponding to the multiple warning levels. For example, according to the number of users at each warning level, the user ratio of each warning level is calculated, and then a pie chart is generated using the warning colors and user ratios of each warning level. The pie chart contains multiple sector areas, each sector area is associated with a user ratio of a warning level, and the ratio of the area of ​​the sector area to the entire pie chart is the user ratio of the warning level associated with the sector area. The area color of the sector area is the warning color of the warning level associated with the sector area.

[0283] For ease of understanding, Figure 5 An example diagram of early warning information is shown. Figure 5 The warning page 500 shown may include the amount of arrears data with a higher arrears level, specific arrears data, collection work orders, supervision work orders, and the like.

[0284] In the embodiment of the present disclosure, when warning the number of users at different warning levels, different warning colors can be assigned to different warning levels, and then warning information is generated according to the warning colors corresponding to each warning level. Each warning level in the warning information is prompted with a different warning color, which can intuitively help people quickly judge the current warning situation and reduce the possibility of misjudgment.

[0285] As yet another embodiment, providing a warning prompt to a user according to the user's target warning level includes:

[0286] Determining a target prompting method corresponding to a target warning level according to the prompting methods corresponding to at least one warning level;

[0287] According to the target prompt mode, generate warning prompt information corresponding to the target warning level;

[0288] Output warning prompt information.

[0289] Optionally, multiple prompting methods may be preset, each prompting method including a prompting template, which may be any type of template such as a text template, a web page template, a text message template, a collection letter template, etc.

[0290] Optionally, each warning level may be associated with a corresponding prompting method, and the prompting method may be corresponding to the warning level. The higher the warning level, the more efficient the prompting method.

[0291] Exemplarily, if the target prompt mode is an email prompt mode, in response to the target prompt mode being an email prompt mode, the user's mailbox is obtained, and an email with the warning prompt information and the email body is sent to the user's mailbox as the inbox.

[0292] Optionally, before outputting the warning prompt information, it also includes: generating a warning confirmation pop-up window based on the warning prompt information, and the pop-up window content of the warning confirmation pop-up window is the warning prompt information. The warning confirmation pop-up window includes a confirmation button and a cancel button. In response to the trigger operation performed on the confirmation button of the warning confirmation pop-up window, the warning prompt information is sent to the user terminal.

[0293] For ease of understanding, Figure 6 An example diagram of an early warning confirmation pop-up window is shown. The early warning confirmation pop-up window 600 may include a display window 601 of early warning prompt information composed of text such as "Dear customer, as of ** month ** day ** year, there are two products under your group's end-to-end business test, and there is still 720 yuan of fees to be settled... You can contact your account manager Xiao A", as well as a confirmation button 602 and a cancel button 603.

[0294] In the disclosed embodiment, when a user is given an early warning prompt according to the user's target early warning level, a target prompt method corresponding to the target early warning level can be used for prompting. The setting of the target prompt method makes the prompt method of each early warning prompt level clearer. For managers and decision makers, timely and efficient prompts can help strengthen risk management and improve the organization's ability to resist risks. By establishing a sound early warning mechanism, organizations can timely discover and deal with potential risks and reduce the probability and impact of risks.

[0295] Figure 7 A schematic diagram of the structure of an early warning analysis device provided in an embodiment of the present disclosure, the early warning analysis device may include the following units:

[0296] The analysis unit 701 is used to perform early warning analysis on the user according to at least one user attribute data of the user to obtain the first credit level and the first characteristic value of the user, wherein each user attribute data includes attribute data corresponding to at least one user attribute of the user.

[0297] The acquisition unit 702 is used to acquire multiple warning areas in the coordinate system. The multiple warning areas are obtained based on the warning curve in the coordinate system. The multiple warning areas are respectively associated with corresponding warning levels. The coordinate system is constructed with the credit and the characteristic value as the coordinate axis.

[0298] The mapping unit 703 is used to map the first credit and the first characteristic value into a coordinate system to obtain a first coordinate point.

[0299] The associating unit 704 is configured to determine the warning level associated with the warning area of ​​the first coordinate point in the coordinate system as the target warning level of the user.

[0300] The prompting unit 705 is used to provide a warning prompt to the user according to the user's target warning level.

[0301] As an embodiment, the analysis unit may include:

[0302] A first analysis module, configured to perform a credit analysis on the user according to at least one user attribute data of the user, and obtain a first credit rating of the user;

[0303] A second analysis module, used to determine target user attribute data of the user according to at least one user attribute data of the user;

[0304] The feature extraction module is used to extract features of the user according to the target user attribute data to obtain the first feature value of the user.

[0305] As yet another embodiment, the first analysis module may include:

[0306] A first calculation submodule, used to calculate the total information entropy of the user according to the attribute data corresponding to at least one user attribute of each user attribute data;

[0307] A second calculation submodule, used to calculate the user's clustering information entropy according to the clustering result of at least one user attribute data;

[0308] The third calculation submodule is used to calculate the difference between the total information entropy and the cluster information entropy to obtain the information entropy gain;

[0309] A distribution determination submodule, used to determine attribute distribution data corresponding to at least one user attribute;

[0310] The ratio calculation submodule is used to calculate the information gain ratio of the user according to the information entropy gain and the attribute distribution data, and determine the information gain ratio of the user as the first credit of the user.

[0311] As another embodiment, the first calculation submodule may be specifically used for:

[0312] Determine at least one attribute data corresponding to each user attribute according to the attribute data corresponding to at least one user attribute of each user attribute data;

[0313] Calculate the information entropy of each user attribute according to at least one attribute data corresponding to each user attribute;

[0314] The information entropy of each user attribute is weighted and summed to obtain the total information entropy.

[0315] As yet another embodiment, the first computing submodule may be specifically configured to perform at least one of the following operations:

[0316] In response to the user attribute being an age attribute, using an age information entropy calculation formula to calculate the age information entropy corresponding to at least one age of the age attribute;

[0317] In response to the user attribute being an arrears amount attribute, using an arrears information entropy calculation formula to calculate arrears information entropy corresponding to at least one arrears amount of the arrears amount attribute;

[0318] In response to the user attribute being an overdue duration attribute, using a duration information entropy calculation formula to calculate duration information entropy corresponding to at least one overdue duration of the overdue duration attribute;

[0319] In response to the user attribute being a credit score attribute, using a score information entropy calculation formula to calculate the credit information entropy corresponding to at least one credit score of the credit score attribute;

[0320] In response to the user attribute being an arrears history attribute, a historical information entropy calculation formula is used to calculate a rating information entropy corresponding to at least one historical evaluation level of the arrears history attribute.

[0321] As yet another embodiment, the second calculation submodule may be specifically used for:

[0322] Determine a clustering result of at least one user attribute data, the clustering result includes clusters corresponding to cluster centers of each user attribute, each cluster contains at least one user attribute data belonging to the same category;

[0323] Calculate the internal information entropy corresponding to at least one user attribute data in each cluster;

[0324] The internal information entropy of each cluster is weighted and summed to obtain the cluster information entropy.

[0325] As another embodiment, the calculation formula of clustering information entropy is as follows:

[0326]

[0327] Among them, H(D) is the clustering information entropy, H(C i ) is the internal information entropy of the i-th user attribute, C i is the number of at least one user attribute data in the i-th cluster, and D is the number of at least one user attribute data.

[0328] As yet another embodiment, it further includes:

[0329] A first determining unit, configured to initialize a cluster center point of each user attribute according to at least one user attribute, and obtain K cluster center points, where K is the number of the at least one user attribute;

[0330] The data clustering unit is used to cluster at least one user attribute data according to K cluster centers and a preset clustering algorithm to obtain a clustering result, wherein the clustering result includes K clusters, and each cluster includes at least one user attribute data belonging to the same category.

[0331] As yet another embodiment, the distribution determination submodule is specifically configured to:

[0332] According to the distribution attribute calculation formula, the attribute distribution data corresponding to at least one user attribute is calculated. The distribution attribute calculation formula is as follows:

[0333]

[0334] Among them, Values(A) is all possible values ​​of user attribute A, D v is a subset of data with value v on attribute A, |D v | is the number of samples in the dataset D, |D v | is the data subset D v The number of samples.

[0335] As yet another embodiment, the feature extraction module includes:

[0336] The feature extraction submodule is used to calculate the first feature value corresponding to the target user attribute data according to the feature extraction formula. The feature extraction formula is as follows:

[0337]

[0338] Among them, Y is the first characteristic value, and X1-X5 are attribute data corresponding to at least one user attribute of the target user attribute data.

[0339] As yet another embodiment, it further includes:

[0340] A coordinate construction unit, used for constructing a coordinate system with the credit and the characteristic value as coordinate axes respectively;

[0341] The area division unit is used to determine the warning curve in the coordinate system based on the preset warning analysis formula. The warning curve is used to divide the coordinate system into multiple warning areas, and each warning area corresponds to a corresponding warning level;

[0342] The mapping unit may include:

[0343] The data mapping module is used to determine the first coordinate point of the first credit and the first characteristic value in the coordinate system according to the credit coordinate axis and the characteristic value coordinate axis of the coordinate system.

[0344] As another embodiment, the area division unit may include:

[0345] A parameter determination module, used to determine a first parameter and a second parameter in a preset early warning analysis formula, wherein the first parameter is a formula used to adjust the algorithm composition in the early warning analysis formula, and the second parameter is an input parameter of the early warning analysis formula;

[0346] A random value module, used for randomly determining a first value of a first parameter and a second value of a second parameter;

[0347] A data input module, used to input a first value of the first parameter and a second value of the second parameter into an early warning analysis formula, and calculate and obtain early warning data corresponding to the first value and the second value;

[0348] A data drawing module is used to draw the second parameter and the warning data as coordinate points in a coordinate system to obtain the coordinate points in the coordinate system;

[0349] The curve fitting module is used to perform curve fitting on multiple coordinate points in the coordinate system to obtain a warning curve.

[0350] As another embodiment, the early warning analysis formula is as follows:

[0351]

[0352] Among them, Y is the second parameter, a, b, c, d, w are all first parameters, and F(Y) is the early warning data obtained by calculation.

[0353] As yet another embodiment, the user includes multiple users, and the device further includes:

[0354] A level division unit, used to divide users with the same target warning level among multiple users into the same user group, and obtain user groups corresponding to each warning level, each user group includes at least one user with the same warning level;

[0355] A quantity determination unit, used to determine the number of users at each warning level according to the user groups corresponding to each warning level;

[0356] The quantity output unit is used to output the number of users corresponding to multiple warning levels.

[0357] As yet another embodiment, the quantity output unit may include:

[0358] A color allocation module, used to allocate warning colors to multiple warning levels respectively;

[0359] The information generation module is used to generate warning information of the number of users corresponding to multiple warning levels according to the warning colors corresponding to each warning level. The warning information displays the ratio of the number of at least one user in the corresponding user group to the number of multiple users in the warning color of each warning level.

[0360] As another embodiment, the prompting unit may include:

[0361] A mode determination module, used to determine a target prompt mode corresponding to a target warning level according to the prompt modes corresponding to at least one warning level;

[0362] A template generation module is used to generate warning prompt information corresponding to the target warning level according to the target prompt mode;

[0363] The information output module is used to output early warning prompt information.

[0364] Figure 8 A schematic diagram of the structure of a computer device provided in an embodiment of the present disclosure is shown. The computer device may include a memory 801, a processor 802, and a computer program stored in the memory. The processor executes the computer program to implement any one of the early warning analysis methods in the above embodiments.

[0365] For example, the processor can be used to perform a warning analysis on the user based on at least one user attribute data of the user, obtain the user's first credit and first eigenvalue, each user attribute data includes attribute data corresponding to at least one user attribute of the user; obtain multiple warning areas in the coordinate system, the multiple warning areas are obtained based on the warning curve in the coordinate system, the multiple warning areas are respectively associated with corresponding warning levels, and the coordinate system is constructed with the credit and the eigenvalue as coordinate axes respectively; map the first credit and the first eigenvalue to the coordinate system to obtain a first coordinate point; determine the warning level associated with the warning area of ​​the first coordinate point in the coordinate system as the user's target warning level; and issue a warning prompt to the user according to the user's target warning level.

[0366] In addition, an embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, any one of the early warning analysis methods in the above embodiments is implemented.

[0367] In addition, an embodiment of the present application also provides a computer program product, including a computer program, which implements any early warning analysis method in the above embodiments when executed by a processor.

[0368] In the above, the early warning analysis method according to the embodiment of the present disclosure is described with reference to the accompanying drawings, a large amount of user attribute data of the user is obtained, and the user attribute data of the user is subjected to early warning analysis, and the first credit and the first characteristic value of the user are obtained, so that the first credit can more accurately characterize the credit of the user, and the first characteristic value can accurately characterize the characteristics of the user. The first credit and the first characteristic value are thus mapped to a coordinate system, and the coordinate system is divided into a plurality of early warning areas. Therefore, the early warning area where the coordinate points of the first credit and the first characteristic value in the coordinate system are located is the final mapping result, and the early warning level associated with the early warning area is the target early warning level of the user, and the target early warning level is used to give an early warning prompt to the user. The early warning level of the user can be obtained by mapping characteristics and credit, and the division of early warning levels can realize early warning identification and monitoring in finer dimensions, which can effectively improve the accuracy of early warnings.

[0369] The basic principles of the present disclosure are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present disclosure are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present disclosure. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, and are not limitations. The above details do not limit the present disclosure to the necessity of adopting the above specific details to be implemented.

[0370] The block diagrams of the devices, apparatuses, equipment, and systems involved in this disclosure are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including," "comprising," "having," and the like are open words, referring to "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or," and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0371] Additionally, as used herein, "or" used in a list of items beginning with "at least one" indicates a separate list, so that, for example, a list of "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not mean that the example described is preferred or better than other examples.

[0372] It should also be noted that in the system and method of the present disclosure, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present disclosure.

[0373] Various changes, substitutions, and modifications of the technology herein may be made without departing from the technology of the teachings defined by the appended claims. Furthermore, the scope of the claims of the present disclosure is not limited to the specific aspects of the above processes, machines, manufactures, compositions of events, means, methods, and actions. Currently existing or later to be developed processes, machines, manufactures, compositions of events, means, methods, or actions that perform substantially the same functions or achieve substantially the same results as the corresponding aspects herein may be utilized. Thus, the appended claims include such processes, machines, manufactures, compositions of events, means, methods, or actions within their scope.

[0374] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0375] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A warning analysis method, characterized in that: include: Performing early warning analysis on the user to obtain a first credit rating and a first characteristic value of the user according to at least one user attribute data of the user, wherein each user attribute data includes attribute data corresponding to at least one user attribute of the user; Acquire multiple warning areas in a coordinate system, the multiple warning areas are obtained based on the division of the warning curve in the coordinate system, the multiple warning areas are respectively associated with corresponding warning levels, and the coordinate system is constructed with the credit and the characteristic value as coordinate axes; Mapping the first credit and the first characteristic value into the coordinate system to obtain a first coordinate point; Determining the warning level associated with the warning area of ​​the first coordinate point in the coordinate system as the target warning level of the user; Providing a warning prompt to the user according to the target warning level of the user.

2. The method according to claim 1, characterized in that The step of performing early warning analysis on the user according to at least one user attribute data of the user to obtain a first credit rating and a first characteristic value of the user includes: Performing a credit analysis on the user according to at least one user attribute data of the user to obtain the first credit level of the user; Determining target user attribute data of the user according to at least one user attribute data of the user; According to the target user attribute data, feature extraction is performed on the user to obtain the first feature value of the user.

3. The method according to claim 2, characterized in that The step of performing a credit analysis on the user according to at least one user attribute data of the user to obtain the first credit level of the user includes: Calculate the total information entropy of the user according to the attribute data corresponding to at least one user attribute of each user attribute data; Calculating the clustering information entropy of the user according to the clustering result of the at least one user attribute data; Calculating the difference between the total information entropy and the cluster information entropy to obtain information entropy gain; Determining attribute distribution data corresponding to the at least one user attribute; The information gain ratio of the user is calculated according to the information entropy gain and the attribute distribution data, and the information gain ratio of the user is determined as the first credit level of the user.

4. The method according to claim 3, characterized in that: The calculating the total information entropy of the user according to the attribute data corresponding to at least one user attribute of each user attribute data includes: Determine at least one attribute data corresponding to each user attribute according to the attribute data corresponding to at least one user attribute of each user attribute data; Calculate the information entropy of each user attribute according to at least one attribute data corresponding to each user attribute; The information entropy of each user attribute is weighted and summed to obtain the total information entropy.

5. The method according to claim 4, characterized in that The calculating the information entropy of each user attribute according to at least one attribute data corresponding to each user attribute includes at least one of the following: In response to the user attribute being an age attribute, using an age information entropy calculation formula to calculate age information entropy corresponding to at least one age of the age attribute; In response to the user attribute being an arrears amount attribute, using an arrears information entropy calculation formula to calculate arrears information entropy corresponding to at least one arrears amount of the arrears amount attribute; In response to the user attribute being an arrears duration attribute, using a duration information entropy calculation formula to calculate duration information entropy corresponding to at least one arrears duration of the arrears duration attribute; In response to the user attribute being a credit score attribute, using a score information entropy calculation formula to calculate the credit information entropy corresponding to at least one credit score of the credit score attribute; In response to the user attribute being an arrears history attribute, a historical information entropy calculation formula is used to calculate a rating information entropy corresponding to at least one historical evaluation level of the arrears history attribute.

6. The method according to claim 3, characterized in that The calculating the clustering information entropy of the user according to the clustering result of the at least one user attribute data comprises: Determine a clustering result of the at least one user attribute data, the clustering result comprising clusters corresponding to cluster centers of each user attribute, each cluster containing at least one user attribute data belonging to the same category; Calculate the internal information entropy corresponding to at least one user attribute data in each cluster; The internal information entropy of each cluster is weighted and summed to obtain the cluster information entropy.

7. The method according to claim 6, characterized in that The calculation formula of the clustering information entropy is as follows: Among them, H(D) is the clustering information entropy, H(C i ) is the internal information entropy of the i-th user attribute, C i is the number of at least one user attribute data in the i-th cluster, and D is the number of the at least one user attribute data.

8. The method according to claim 3, characterized in that Also includes: Initializing a cluster center point of each user attribute according to the at least one user attribute to obtain K cluster center points, where K is the number of the at least one user attribute; According to the K cluster centers and a preset clustering algorithm, the at least one user attribute data is clustered to obtain the clustering result, wherein the clustering result includes K clusters, and each cluster includes at least one user attribute data belonging to the same category.

9. The method according to claim 3, characterized in that: The determining the attribute distribution data corresponding to the at least one user attribute includes: Calculate the attribute distribution data corresponding to the at least one user attribute according to the distribution attribute calculation formula, the distribution attribute calculation formula is as follows: Among them, Values(A) is all possible values ​​of user attribute A, D v is a subset of data with value v on attribute A, |D v | is the number of samples in the dataset D, |D v | is the data subset D v The number of samples.

10. The method according to claim 2, characterized in that The step of extracting features of the user according to the target user attribute data to obtain the first feature value of the user includes: According to the feature extraction formula, the first feature value corresponding to the target user attribute data is calculated. The feature extraction formula is as follows: Among them, Y is the first characteristic value, and X1-X5 are the attribute data corresponding to at least one user attribute of the target user attribute data.

11. The method according to any one of claims 1 to 10, characterized in that: Also includes: Constructing the coordinate system with the credit and the characteristic value as coordinate axes respectively; Based on a preset warning analysis formula, a warning curve in the coordinate system is determined, wherein the warning curve is used to divide the coordinate system into a plurality of warning areas, each warning area corresponding to a corresponding warning level; Mapping the first credit and the first characteristic value into the coordinate system to obtain a first coordinate point includes: According to the credit coordinate axis and the characteristic value coordinate axis of the coordinate system, the first coordinate points of the first credit and the first characteristic value in the coordinate system are determined.

12. The method according to claim 11, characterized in that The early warning curve in the coordinate system is determined based on the preset early warning analysis formula, including: Determine a first parameter and a second parameter in the preset early warning analysis formula, wherein the first parameter is a formula in the early warning analysis formula for adjusting the algorithm composition, and the second parameter is an input parameter of the early warning analysis formula; Randomly determine a first value of the first parameter and a second value of the second parameter; Inputting a first value of the first parameter and a second value of the second parameter into the early warning analysis formula, and calculating and obtaining early warning data corresponding to the first value and the second value; Plotting the second parameter and the warning data as coordinate points in the coordinate system to obtain coordinate points in the coordinate system; A plurality of coordinate points in the coordinate system are subjected to curve fitting to obtain the warning curve.

13. The method according to claim 12, characterized in that The early warning analysis formula is as follows: Among them, Y is the second parameter, a, b, c, d, w are all first parameters, and F(Y) is the early warning data obtained by calculation.

14. The method according to any one of claims 1 to 10, characterized in that: The user includes multiple users, and the method further includes: Divide users with the same target warning level among multiple users into the same user group to obtain user groups corresponding to each warning level, each user group including at least one user with the same warning level; Determine the number of users at each warning level based on the user groups corresponding to each warning level; Output the number of users corresponding to each of the plurality of warning levels.

15. The method according to claim 14, characterized in that The outputting the number of users corresponding to the plurality of warning levels respectively comprises: Assign warning colors to the multiple warning levels respectively; According to the warning colors corresponding to the warning levels, warning information of the number of users corresponding to the multiple warning levels is generated, and the warning information displays the ratio of the number of at least one user in the corresponding user group to the number of multiple users in the warning color of each warning level.

16. The method according to any one of claims 1 to 10, characterized in that: Providing a warning prompt to the user according to the target warning level of the user includes: Determining a target prompting mode corresponding to the target warning level according to the prompting modes corresponding to the at least one warning level; According to the target prompt mode, generating warning prompt information corresponding to the target warning level; Output the warning prompt information.

17. An early warning analysis device, characterized in that: include: An analysis unit, configured to perform early warning analysis on the user to obtain a first credit rating and a first characteristic value of the user according to at least one user attribute data of the user, wherein each user attribute data includes attribute data corresponding to at least one user attribute of the user; An acquisition unit, used for acquiring a plurality of warning areas in a coordinate system, wherein the plurality of warning areas are obtained based on the division of the warning curve in the coordinate system, the plurality of warning areas are respectively associated with corresponding warning levels, and the coordinate system is respectively constructed with the credit and the characteristic value as coordinate axes; A mapping unit, used for mapping the first credit and the first characteristic value into the coordinate system to obtain a first coordinate point; an associating unit, configured to determine the warning level associated with the warning area of ​​the first coordinate point in the coordinate system as the target warning level of the user; A prompting unit is used to provide a warning prompt to the user according to the target warning level of the user.

18. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 16.

19. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to claims 1 to 16 are implemented.

20. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to claims 1 to 16 are implemented.

Images