CN117132001B - Multi-target wind control strategy optimization method and system - Google Patents
Multi-target wind control strategy optimization method and system Download PDFInfo
- Publication number
- CN117132001B CN117132001B CN202311379683.1A CN202311379683A CN117132001B CN 117132001 B CN117132001 B CN 117132001B CN 202311379683 A CN202311379683 A CN 202311379683A CN 117132001 B CN117132001 B CN 117132001B
- Authority
- CN
- China
- Prior art keywords
- data
- determining
- risk
- credit application
- input
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000005457 optimization Methods 0.000 title claims abstract description 21
- 238000011217 control strategy Methods 0.000 title claims abstract description 20
- 238000011156 evaluation Methods 0.000 claims abstract description 81
- 238000012545 processing Methods 0.000 claims abstract description 17
- 238000012502 risk assessment Methods 0.000 claims description 18
- 238000013075 data extraction Methods 0.000 claims description 12
- 238000007477 logistic regression Methods 0.000 claims description 4
- 238000012417 linear regression Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 description 6
- 230000007774 longterm Effects 0.000 description 4
- 238000010276 construction Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- Finance (AREA)
- Human Resources & Organizations (AREA)
- Accounting & Taxation (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Technology Law (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention provides a multi-target wind control strategy optimization method and a system, which belong to the technical field of data processing and specifically comprise the following steps: the method comprises the steps of determining the association coefficient of the credit application data and the wind control result according to the credit application data of a user, determining comprehensive evaluation values and input data of the credit application data of different data types based on the association coefficient, the data quantity and the data source of the credit application data of different data types, determining the data type and the quantity of the credit application data in the model input data of the wind control model based on an interpretability threshold and the comprehensive evaluation values, and determining the credit application risk of the user by combining the initial evaluation overdue risk and the wind control model, so that the interpretability and the wind control processing efficiency of the wind control model are further improved.
Description
Technical Field
The invention belongs to the technical field of data processing, and particularly relates to a multi-target wind control strategy optimization method and system.
Background
In order to realize identification and processing of the trust risk of the user, the prior art scheme often realizes identification of the trust risk of the user through construction based on the wind control model, but how to realize balance between the processing efficiency and the accuracy of the wind control model becomes a technical problem to be solved urgently.
In order to realize the balance between the processing efficiency and the accuracy of the wind control model, in the invention patent CN202110855572. X & lt- & gt, a wind control model construction method, a device and electronic equipment based on multi-objective optimization, an initial wind control model comprising a plurality of cascade logistic regression algorithms is constructed, and multi-objective combined training is carried out on the initial wind control model according to a service sample training set, so that a final wind control model is obtained, and the accurate assessment of wind control risks is realized, but the following technical problems exist:
in the prior art, evaluation of the interpretability of the wind control model is ignored, and in the process of building the wind control model, the simple adoption of a regression model or a deep learning model cannot ensure the balance of the interpretability and the accuracy of the whole wind control model, and meanwhile, if the interpretability of the wind control model is ignored, the requirements of a supervision department cannot be met, and meanwhile, the problem of the wind control model cannot be solved.
In order to solve the technical problems, the invention provides a multi-target wind control strategy optimization method and a system.
Disclosure of Invention
In order to achieve the purpose of the invention, the invention adopts the following technical scheme:
according to one aspect of the invention, a multi-objective wind control strategy optimization method is provided.
The multi-target wind control strategy optimization method is characterized by comprising the following steps of:
s1, determining association coefficients of the credit application data and an air control result according to credit application data of a user, and performing comprehensive evaluation of the credit application data of different data types and determination of input data based on the association coefficients, data quantity and data sources of the credit application data of different data types;
s2, dividing an input data cluster according to the data types of the input data, obtaining initial evaluation overdue risks of users under different data types based on the data types of different input data clusters and the regression models corresponding to the data types, judging whether the initial evaluation overdue risks larger than the preset risks exist for the users, if so, entering a step S3, otherwise, entering a step S4;
s3, acquiring the number of the input data clusters of the user, determining the initial comprehensive overdue risk of the user by combining the data sources of the input data of different data types of the user and the initial evaluation overdue risk, and entering the next step when the initial comprehensive overdue risk meets the requirement;
and S4, determining the data type and the number of the trust application data in the model input data of the wind control model based on the interpretability threshold and the comprehensive evaluation quantity, and determining the trust risk of the user by combining the initial evaluation overdue risk and the wind control model.
The invention has the beneficial effects that:
1. the comprehensive evaluation of the trust application data of different data types and the determination of the input data are carried out based on the association coefficients, the data quantity and the data sources of the trust application data of different data types, so that the difference of the association coefficients of the trust application data of the data types which do not pass through is considered, the authenticity and the extraction difficulty are considered, the association, the authenticity and the extraction difficulty of the data are all within a certain range, and the processing accuracy and the timeliness of the trust risk are ensured.
2. The initial evaluation overdue risk of the user under different data types is obtained based on the data types of different input data clusters and the regression models corresponding to the data types, so that the interpretability of the model is greatly improved, and the technical problem of insufficient interpretability caused by single machine learning algorithm is avoided.
3. By determining the initial comprehensive overdue risk of the user, the overdue risk of initial evaluation of different data types is simply considered, meanwhile, the accuracy and the reliability of an evaluation result are further evaluated from the perspective of data sources, the comprehensive evaluation of the overdue risk of the user is realized by comprehensively considering the number of input data clusters, and the screening of the user with higher overdue risk is also realized.
4. The data type and the number of the trust application data in the model input data of the wind control model are determined based on the interpretability threshold and the comprehensive evaluation quantity, so that the interpretability of the final wind control model is considered, and meanwhile, the accuracy and the reliability of the final wind control model are considered, and the overdue risk evaluation of a user can be accurately and reliably realized.
The further technical scheme is that the trust application data of the user is determined according to the filling data of the trust application of the user and the authorized access information data of the user.
The further technical scheme is that the correlation coefficient of the credit application data and the wind control result is determined according to the correlation of the credit application data and the wind control result at different times, and particularly, the correlation coefficient of the credit application data and the wind control result at different times is determined.
The further technical scheme is that when the comprehensive evaluation of the data type credit application data is within a specified data range, the data type credit application data is determined to be input data.
The further technical scheme is that the regression model is built through a Logistic regression model or a linear regression model.
The further technical scheme is that the preset risk is determined according to the application amount of the credit application users of the credit application processing enterprises and the average application amount in unit time, wherein the larger the application amount of the credit application users of the credit application processing enterprises is, the more the average application amount in unit time is, and the lower the preset risk is.
The further technical scheme is that the method for determining the data type and the number of the credit application data in the model input data of the wind control model based on the interpretability threshold and the comprehensive evaluation quantity specifically comprises the following steps:
determining the interpretability of the model input data based on the ratio of the number of the credit application data in the model input data to the number of the model input data, and determining the number of the credit application data in the model input data by combining the interpretability threshold;
and determining the data type of the credit application data in the model input data of the wind control model according to the number of the credit application data and the comprehensive evaluation quantity.
In a second aspect, the present invention provides a multi-objective wind control strategy optimization system, and the multi-objective wind control strategy optimization method is adopted, and the method is characterized in that the method specifically includes:
the system comprises an input data determining module, an initial risk assessment module, a comprehensive risk assessment module and a credit risk assessment module;
the input data determining module is responsible for determining the association coefficient of the credit application data and the wind control result according to the credit application data of the user, and performing comprehensive evaluation of the credit application data of different data types and determination of the input data based on the association coefficient, the data quantity and the data source of the credit application data of different data types;
the initial risk assessment module is responsible for dividing input data clusters according to the data types of the input data, and obtaining initial assessment overdue risks of users under different data types based on the data types of different input data clusters and regression models corresponding to the data types;
the comprehensive risk assessment module is responsible for acquiring the number of input data clusters of the user, and determining the initial comprehensive overdue risk of the user by combining data sources of input data of different data types of the user and the initial assessment overdue risk;
the trust risk assessment module is responsible for determining the data type and the number of trust application data in the model input data of the wind control model based on the interpretability threshold and the comprehensive assessment amount, and determining the trust risk of the user by combining the initial assessment overdue risk and the wind control model.
Additional features and advantages will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and drawings.
In order to make the above objects, features and advantages of the present invention more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
The above and other features and advantages of the present invention will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings;
FIG. 1 is a flow chart of a multi-objective wind control strategy optimization method;
FIG. 2 is a flowchart showing specific steps for determining the comprehensive evaluation amount of the credit application data of different data types;
FIG. 3 is a flowchart showing specific steps for determining an initial assessment of a comprehensive overdue risk for a user;
FIG. 4 is a framework diagram of a multi-objective wind control strategy optimization system.
Detailed Description
In order to make the technical solutions in the present specification better understood by those skilled in the art, the technical solutions in the embodiments of the present specification will be clearly and completely described below with reference to the drawings in the embodiments of the present specification, and it is obvious that the described embodiments are only some embodiments of the present specification, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, shall fall within the scope of the present disclosure.
The applicant finds that in the prior art, when the wind control model is built, the interpretability and the reliability of the wind control model are not considered at the same time, so that the requirement of supervision cannot be met, the interpretability of the model is poor, and the targeted improvement treatment of the model cannot be accurately realized once a problem occurs.
In order to solve the technical problems, the applicant adopts the following technical scheme:
firstly, determining comprehensive evaluation quantity of different types of credit application data through association coefficients, data quantity and data sources of different data types and wind control results, specifically, determining extraction difficulty through the data quantity, evaluating credibility through the quantity of the data sources, determining comprehensive evaluation quantity according to products of the association coefficients, the inverse of the extraction difficulty and the credibility of the credit application data of different data types and wind control results, and taking credit application data with larger comprehensive evaluation quantity as input data;
setting a corresponding regression model for different types of input data according to the different types of input data, so as to obtain initial evaluation overdue risk under the different types of input data, determining the initial overdue risk of the user under the different types of input data according to the initial evaluation overdue risk of the user under the different types of input data and the number of data sources of the different types of input data, determining the initial overdue risk sum by combining the number of the different types of input data of the user, dividing the initial overdue risk sum by the number of the different types of input data to obtain the initial comprehensive overdue risk of the user, directly outputting the processing that cannot carry out credit application at the moment when the initial comprehensive overdue risk of the user is larger, leading the risk to be higher, and entering the next step when the initial comprehensive overdue risk of the user is smaller;
determining the interpretability of the model input data based on the ratio of the number of the credit application data in the model input data to the number of the model input data, and determining the number of the credit application data in the model input data by combining a preset interpretability threshold; determining the data type of the credit application data in the model input data of the wind control model according to the quantity and the comprehensive evaluation quantity of the credit application data;
and determining the model input data of the wind control model according to the initial evaluation overdue risk and the data type and the number of the trust application data in the model input data, and determining the trust risk of the user according to the model input data of the wind control model and the wind control model.
The following is a detailed description from both the perspective of the method class embodiment and the system class embodiment.
In order to solve the above-mentioned problems, according to an aspect of the present invention, as shown in fig. 1, there is provided a multi-objective wind control strategy optimization method, which is characterized by comprising:
s1, determining association coefficients of the credit application data and an air control result according to credit application data of a user, and performing comprehensive evaluation of the credit application data of different data types and determination of input data based on the association coefficients, data quantity and data sources of the credit application data of different data types;
the user's credit application data is determined according to the filling data of the user's credit application and the user's authorized access information data.
Specifically, the correlation coefficient between the credit application data and the wind control result is determined according to the correlation between the credit application data at different times and the wind control result, and specifically, the correlation coefficient between the credit application data at different times and the wind control result is determined.
In one possible embodiment, the specific step of determining the association coefficient between the trust application data and the wind control result in the step S1 is:
dividing the credit application data of the history application user into recent credit application data and long-term credit application data based on preset time length, evaluating correlation coefficients of the recent credit application data and long-term credit application data of the history application user and the wind control result of the history application user to obtain recent correlation coefficients and long-term correlation coefficients of the credit application data, and determining correlation coefficients of the credit application data of the history application user and the wind control result according to the recent correlation coefficients and the long-term correlation coefficients;
when the correlation coefficients of the trust application data of the historical application user and the wind control result are all larger than a preset value:
determining the correlation coefficient of the credit application data and the wind control result based on the average value of the correlation coefficient of the credit application data and the wind control result of all the history application users and the number of the history application users;
when the association coefficient between the credit application data of any one historical application user and the wind control result is not larger than a preset value:
judging whether the number of history application users with the association coefficient of the credit application data and the wind control result not larger than a preset value and the ratio of the history application users in the number of all the history application users meet the requirements, if so, determining the association coefficient of the credit application data and the wind control result based on the average value of the association coefficients of the credit application data and the wind control result of all the history application users and the number of the history application users, and if not, entering the next step;
and taking the history application user with the correlation coefficient of the credit application data and the wind control result not larger than a preset value as a low correlation user, and determining the correlation coefficient of the credit application data and the wind control result through the number of the low correlation users, the average value of the correlation coefficient, the average value of the credit application data of the history application user and the correlation coefficient of the wind control result and the number of the history application user.
In one possible embodiment, as shown in fig. 2, the specific steps of determining the comprehensive evaluation value of the trust application data of the different data types in the step S1 are as follows:
s11, determining whether the data type of the credit application data belongs to input data or not based on the association coefficient of the data type of the credit application data, if so, determining the data type of the credit application data as the input data, determining the comprehensive evaluation amount of the data type of the credit application data based on the association coefficient, and if not, entering the next step;
s12, judging whether the association coefficient of the data type of the trust application data is larger than a preset association value, if so, entering the next step, and if not, entering the step S15;
s13, determining the data extraction difficulty of the data type credit application data based on the data quantity of the data type credit application data, determining whether the data type credit application data belongs to input data or not according to the data extraction difficulty of the data type credit application data, if so, determining the data type credit application data as the input data, and determining the comprehensive evaluation quantity of the data type credit application data based on the association coefficient and the data extraction difficulty, otherwise, entering the next step;
s14, determining the type and the number of the data sources of the data type credit application data based on the data sources of the data type credit application data, determining the credibility of the data type credit application data by combining the update time of the data sources of the data type credit application data, determining whether the data type credit application data belongs to input data or not based on the credibility, if so, determining the data type credit application data as the input data, and determining the comprehensive evaluation amount of the data type credit application data based on the association coefficient and the credibility, otherwise, entering the next step;
s15, determining the comprehensive evaluation quantity of the data type credit application data through the credibility, the data extraction difficulty and the association coefficient of the data type credit application data.
It should be noted that, in the step S1, when the comprehensive evaluation value of the data type of the trust application data is within the specified data range, the trust application data of the data type is determined as the input data.
In another possible embodiment, the specific steps of the comprehensive evaluation determination of the credit application data in the step S1 are:
determining the data extraction difficulty of the data type credit application data based on the data quantity of the data type credit application data, determining the type and the quantity of the data source of the data type credit application data based on the data source of the data type credit application data, and determining the credibility of the data type credit application data by combining the update time of the data source of the data type credit application data;
acquiring the association coefficient of the credit application data of the data type, and when the association coefficient of the credit application data of the data type is larger than a preset association value:
when any one of the credibility of the data type of the trust application data or the data extraction difficulty meets the requirement:
determining the credit application data of the data type as input data, and determining the comprehensive evaluation quantity of the credit application data of the data type through the association coefficient of the credit application data of the data type;
when the association coefficient of the data type credit application data is not larger than a preset association value or the credibility or the data extraction difficulty of the data type credit application data does not meet the requirements:
and determining the comprehensive evaluation quantity of the data type credit application data through the credibility, the data extraction difficulty and the association coefficient of the data type credit application data.
In this embodiment, by performing comprehensive evaluation of the trust application data of different data types and determination of the input data based on the association coefficients, the data amounts and the data sources of the trust application data of different data types, not only the difference of the association coefficients of the trust application data of the data types which do not pass is considered, but also the authenticity and the extraction difficulty are considered, so that the relevance, the authenticity and the extraction difficulty of the data are all within a certain range, and the processing accuracy and the timeliness of the trust risk are ensured.
S2, dividing an input data cluster according to the data types of the input data, obtaining initial evaluation overdue risks of users under different data types based on the data types of different input data clusters and the regression models corresponding to the data types, judging whether the initial evaluation overdue risks larger than the preset risks exist for the users, if so, entering a step S3, otherwise, entering a step S4;
in one possible embodiment, the regression model in the step S2 is built by using a Logistic regression model or a linear regression model.
Specifically, the preset risk in the step S2 is determined according to the application amount of the trusted application user of the trusted application processing enterprise and the average application amount in the unit time, where the larger the application amount of the trusted application user of the trusted application processing enterprise is, the more the average application amount in the unit time is, the lower the preset risk is.
In this embodiment, the initial evaluation overdue risk of the user under different data types is obtained based on the data types of different input data clusters and the regression models corresponding to the data types, so that the interpretability of the model is greatly improved, and the technical problem of insufficient interpretability caused by a single machine learning algorithm is avoided.
S3, acquiring the number of the input data clusters of the user, determining the initial comprehensive overdue risk of the user by combining the data sources of the input data of different data types of the user and the initial evaluation overdue risk, and entering the next step when the initial comprehensive overdue risk meets the requirement;
in one possible embodiment, as shown in fig. 3, the specific steps of determining the initial evaluation comprehensive overdue risk of the user in the step S3 are as follows:
s31, taking input data of a data type with the initial evaluation overdue risk greater than a preset risk as input risk data, determining the type and the number of the data sources of the input risk data through the data sources of the input risk data, determining the credibility of the input risk data by combining the update time of the data sources of the input risk data, judging whether the input risk data with the credibility greater than the preset credibility exists, if so, determining that the initial comprehensive overdue risk does not meet the requirement, and determining that the credibility of the user is higher, otherwise, entering the next step;
s32, determining whether the user has overdue risk or not according to the maximum value of the credibility of the input risk data, if so, entering the next step, and if not, entering the step S34;
s33, determining the maximum value and the average value of the reliability of the input risk data through the reliability of the input risk data, determining the comprehensive reliability of the input risk data by combining the quantity that the reliability of the input risk data is larger than the average value of the reliability, determining whether the user has overdue risk or not based on the comprehensive reliability, if so, determining that the initial comprehensive overdue risk does not meet the requirement, determining that the trusted risk of the user is higher, and if not, entering the next step;
s34, determining initial overdue risks of different input risk data of the user according to the credibility of the input risk data of the user and the initial evaluation overdue risks, and determining initial comprehensive overdue risks of the input risk data by combining the number of the input risk data and the maximum value of the initial overdue risks;
s35, acquiring the number of the input data clusters of the user and initial evaluation overdue risks of different data types, and determining the initial comprehensive overdue risk of the user by combining the initial comprehensive overdue risks of the input risk data of the user.
It should be noted that, in the step S3, when the initial comprehensive overdue risk does not meet the requirement, it is determined that the trust risk of the user is higher, and the trust application cannot be performed.
In another possible embodiment, the specific steps of determining the comprehensive overdue risk of initial evaluation of the user in the step S3 are as follows:
taking input data of a data type with the initial evaluation overdue risk being greater than a preset risk as input risk data, determining the type and the number of the data sources of the input risk data through the data sources of the input risk data, determining the credibility of the input risk data by combining the update time of the data sources of the input risk data, and determining the initial overdue risk of different types of input risk data of the user through the credibility of the input risk data of the user and the initial evaluation overdue risk;
when the user has input risk data with initial overdue risk greater than a preset overdue risk value, the user is provided with input risk data with initial overdue risk greater than the preset overdue risk value:
determining the initial evaluation comprehensive overdue risk of the user according to the quantity of the input risk data with the initial overdue risk being larger than a preset overdue risk value;
when the user does not have input risk data with the initial overdue risk being larger than a preset overdue risk value, the user is provided with the input risk data with the initial overdue risk being larger than the preset overdue risk value:
determining the maximum value and the average value of the credibility of the input risk data according to the credibility of the input risk data, determining the comprehensive credibility of the input risk data according to the number of the input risk data, wherein the credibility of the input risk data is larger than the average value of the credibility, determining whether the user has overdue risk or not based on the comprehensive credibility, if yes, determining that the initial comprehensive overdue risk does not meet the requirement, determining that the credibility risk of the user is higher, and if not, entering the next step;
determining initial overdue risks of different input risk data of the user according to the credibility of the input risk data of the user and the initial evaluation overdue risks, and determining initial comprehensive overdue risks of the input risk data by combining the number of the input risk data and the maximum value of the initial overdue risks; and acquiring the number of the input data clusters of the user and initial evaluation overdue risks of different data types, and determining the initial comprehensive overdue risk of the user by combining the initial comprehensive overdue risks of the input risk data of the user.
In this embodiment, by determining the initial comprehensive overdue risk of the user, not only is the initial overdue risk estimated by considering different data types purely, but also the accuracy and the reliability of the estimation result are further estimated from the perspective of data sources, and the comprehensive estimation of the overdue risk of the user is realized by comprehensively considering the number of input data clusters, and the screening of the user with higher overdue risk is also realized.
And S4, determining the data type and the number of the trust application data in the model input data of the wind control model based on the interpretability threshold and the comprehensive evaluation quantity, and determining the trust risk of the user by combining the initial evaluation overdue risk and the wind control model.
Specifically, the determining of the data type and the number of the trust application data in the model input data of the wind control model based on the interpretability threshold and the comprehensive evaluation amount in the step S4 specifically includes:
determining the interpretability of the model input data based on the ratio of the number of the credit application data in the model input data to the number of the model input data, and determining the number of the credit application data in the model input data by combining the interpretability threshold;
and determining the data type of the credit application data in the model input data of the wind control model according to the number of the credit application data and the comprehensive evaluation quantity.
It should be noted that, in the step S4, the determining the trust risk of the user by combining the initial evaluation overdue risk and the wind control model specifically includes:
and determining the model input data of the wind control model according to the initial evaluation overdue risk and the data type and the number of the trust application data in the model input data, and determining the trust risk of the user according to the model input data of the wind control model and the wind control model.
In this embodiment, the determination of the data type and the number of the trusted application data in the model input data of the wind control model is performed based on the threshold of the interpretability and the comprehensive evaluation amount, so that the interpretability of the final wind control model is considered, and the accuracy and the reliability of the final wind control model are considered, so that the overdue risk of the user can be accurately and reliably evaluated.
On the other hand, as shown in fig. 4, the present invention provides a multi-objective wind control strategy optimization system, and the multi-objective wind control strategy optimization method is adopted, and is characterized in that the method specifically includes:
the system comprises an input data determining module, an initial risk assessment module, a comprehensive risk assessment module and a credit risk assessment module;
the input data determining module is responsible for determining the association coefficient of the credit application data and the wind control result according to the credit application data of the user, and performing comprehensive evaluation of the credit application data of different data types and determination of the input data based on the association coefficient, the data quantity and the data source of the credit application data of different data types;
the initial risk assessment module is responsible for dividing input data clusters according to the data types of the input data, and obtaining initial assessment overdue risks of users under different data types based on the data types of different input data clusters and regression models corresponding to the data types;
the comprehensive risk assessment module is responsible for acquiring the number of input data clusters of the user, and determining the initial comprehensive overdue risk of the user by combining data sources of input data of different data types of the user and the initial assessment overdue risk;
the trust risk assessment module is responsible for determining the data type and the number of trust application data in the model input data of the wind control model based on the interpretability threshold and the comprehensive assessment amount, and determining the trust risk of the user by combining the initial assessment overdue risk and the wind control model.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for apparatus, devices, non-volatile computer storage medium embodiments, the description is relatively simple, as it is substantially similar to method embodiments, with reference to the section of the method embodiments being relevant.
The foregoing describes specific embodiments of the present disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
The foregoing is merely one or more embodiments of the present description and is not intended to limit the present description. Various modifications and alterations to one or more embodiments of this description will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, or the like, which is within the spirit and principles of one or more embodiments of the present description, is intended to be included within the scope of the claims of the present description.
Claims (5)
1. The multi-target wind control strategy optimization method is characterized by comprising the following steps of:
s1, determining association coefficients of the credit application data and an air control result according to credit application data of a user, and performing comprehensive evaluation of the credit application data of different data types and determination of input data based on the association coefficients, data quantity and data sources of the credit application data of different data types;
the correlation coefficient of the credit application data and the wind control result is determined according to the correlation of the credit application data and the wind control result at different times, and particularly, the correlation coefficient of the credit application data and the wind control result at different times is determined;
the specific steps of determining the association coefficient of the trust application data and the wind control result are as follows:
dividing the credit application data of a history application user into recent credit application data and remote credit application data based on preset time length, evaluating correlation coefficients of the recent credit application data and remote credit application data of the history application user and a wind control result of the history application user to obtain recent correlation coefficients and remote correlation coefficients of the credit application data, and determining correlation coefficients of the credit application data of the history application user and the wind control result according to the recent correlation coefficients and remote correlation coefficients;
when the correlation coefficients of the trust application data of the historical application user and the wind control result are all larger than a preset value:
determining the correlation coefficient of the credit application data and the wind control result based on the average value of the correlation coefficient of the credit application data and the wind control result of all the history application users and the number of the history application users;
when the association coefficient between the credit application data of any one historical application user and the wind control result is not larger than a preset value:
judging whether the number of history application users with the association coefficient of the credit application data and the wind control result not larger than a preset value and the ratio of the history application users in the number of all the history application users meet the requirements, if so, determining the association coefficient of the credit application data and the wind control result based on the average value of the association coefficients of the credit application data and the wind control result of all the history application users and the number of the history application users, and if not, entering the next step;
the historical application users with the correlation coefficients of the trust application data and the wind control result not larger than a preset value are used as low correlation users, and the correlation coefficients of the trust application data and the wind control result are determined through the number of the low correlation users, the average value of the correlation coefficients of the trust application data of the historical application users and the wind control result and the number of the historical application users;
the specific steps of comprehensive evaluation and determination of the trust application data of different data types are as follows:
s11, determining whether the data type of the credit application data belongs to input data or not based on the association coefficient of the data type of the credit application data, if so, determining the data type of the credit application data as the input data, determining the comprehensive evaluation amount of the data type of the credit application data based on the association coefficient, and if not, entering the next step;
s12, judging whether the association coefficient of the data type of the trust application data is larger than a preset association value, if so, entering the next step, and if not, entering the step S15;
s13, determining the data extraction difficulty of the data type credit application data based on the data quantity of the data type credit application data, determining whether the data type credit application data belongs to input data or not according to the data extraction difficulty of the data type credit application data, if so, determining the data type credit application data as the input data, and determining the comprehensive evaluation quantity of the data type credit application data based on the association coefficient and the data extraction difficulty, otherwise, entering the next step;
s14, determining the type and the number of the data sources of the data type credit application data based on the data sources of the data type credit application data, determining the credibility of the data type credit application data by combining the update time of the data sources of the data type credit application data, determining whether the data type credit application data belongs to input data or not based on the credibility, if so, determining the data type credit application data as the input data, and determining the comprehensive evaluation amount of the data type credit application data based on the association coefficient and the credibility, otherwise, entering the next step;
s15, determining the comprehensive evaluation quantity of the data type credit application data through the credibility, the data extraction difficulty and the association coefficient of the data type credit application data;
when the comprehensive evaluation value of the data type credit application data is within a specified data range, determining the data type credit application data as input data;
s2, dividing an input data cluster according to the data types of the input data, obtaining initial evaluation overdue risks of users under different data types based on the data types of different input data clusters and the regression models corresponding to the data types, judging whether the initial evaluation overdue risks larger than the preset risks exist for the users, if so, entering a step S3, otherwise, entering a step S4;
the preset risk is determined according to the application amount of the credit application users of the credit application processing enterprises and the average application amount in unit time, wherein the larger the application amount of the credit application users of the credit application processing enterprises is, the more the average application amount in unit time is, the lower the preset risk is;
s3, acquiring the number of the input data clusters of the user, determining the initial comprehensive overdue risk of the user by combining the data sources of the input data of different data types of the user and the initial evaluation overdue risk, and entering the next step when the initial comprehensive overdue risk meets the requirement;
the specific steps of the determination of the initial evaluation comprehensive overdue risk of the user are as follows:
s31, taking input data of a data type with the initial evaluation overdue risk greater than a preset risk as input risk data, determining the type and the number of the data sources of the input risk data through the data sources of the input risk data, determining the credibility of the input risk data by combining the update time of the data sources of the input risk data, judging whether the input risk data with the credibility greater than the preset credibility exists, if so, determining that the initial comprehensive overdue risk does not meet the requirement, and determining that the credibility of the user is higher, otherwise, entering the next step;
s32, determining whether the user has overdue risk or not according to the maximum value of the credibility of the input risk data, if so, entering the next step, and if not, entering the step S34;
s33, determining the maximum value and the average value of the reliability of the input risk data through the reliability of the input risk data, determining the comprehensive reliability of the input risk data by combining the quantity that the reliability of the input risk data is larger than the average value of the reliability, determining whether the user has overdue risk or not based on the comprehensive reliability, if so, determining that the initial comprehensive overdue risk does not meet the requirement, determining that the trusted risk of the user is higher, and if not, entering the next step;
s34, determining initial overdue risks of different input risk data of the user according to the credibility of the input risk data of the user and the initial evaluation overdue risks, and determining initial comprehensive overdue risks of the input risk data by combining the number of the input risk data and the maximum value of the initial overdue risks;
s35, acquiring the number of the input data clusters of the user and initial evaluation overdue risks of different data types, and determining the initial comprehensive overdue risk of the user by combining the initial comprehensive overdue risks of the input risk data of the user;
s4, determining the data type and the number of the trust application data in the model input data of the wind control model based on the interpretability threshold and the comprehensive evaluation quantity, and determining the trust risk of the user by combining the initial evaluation overdue risk and the wind control model;
the method for determining the data type and the number of the credit application data in the model input data of the wind control model based on the interpretability threshold and the comprehensive evaluation quantity specifically comprises the following steps:
determining the interpretability of the model input data based on the ratio of the number of the credit application data in the model input data to the number of the model input data, and determining the number of the credit application data in the model input data by combining the interpretability threshold;
and determining the data type of the credit application data in the model input data of the wind control model according to the number of the credit application data and the comprehensive evaluation quantity.
2. The multi-objective wind control strategy optimization method according to claim 1, wherein the trust application data of the user is determined according to the filling data of the trust application of the user and the authorized access information data of the user.
3. The multi-objective wind control strategy optimization method of claim 1, wherein the regression model is built by a Logistic regression model or a linear regression model.
4. The multi-objective wind control strategy optimization method according to claim 1, wherein when the initial comprehensive overdue risk does not meet the requirement, it is determined that the trust risk of the user is higher, and trust application processing cannot be performed.
5. A multi-objective wind control strategy optimization system, employing a multi-objective wind control strategy optimization method according to any one of claims 1-4, comprising:
the system comprises an input data determining module, an initial risk assessment module, a comprehensive risk assessment module and a credit risk assessment module;
the input data determining module is responsible for determining the association coefficient of the credit application data and the wind control result according to the credit application data of the user, and performing comprehensive evaluation of the credit application data of different data types and determination of the input data based on the association coefficient, the data quantity and the data source of the credit application data of different data types;
the initial risk assessment module is responsible for dividing input data clusters according to the data types of the input data, and obtaining initial assessment overdue risks of users under different data types based on the data types of different input data clusters and regression models corresponding to the data types;
the comprehensive risk assessment module is responsible for acquiring the number of input data clusters of the user, and determining the initial comprehensive overdue risk of the user by combining data sources of input data of different data types of the user and the initial assessment overdue risk;
the trust risk assessment module is responsible for determining the data type and the number of trust application data in the model input data of the wind control model based on the interpretability threshold and the comprehensive assessment amount, and determining the trust risk of the user by combining the initial assessment overdue risk and the wind control model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311379683.1A CN117132001B (en) | 2023-10-24 | 2023-10-24 | Multi-target wind control strategy optimization method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311379683.1A CN117132001B (en) | 2023-10-24 | 2023-10-24 | Multi-target wind control strategy optimization method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117132001A CN117132001A (en) | 2023-11-28 |
| CN117132001B true CN117132001B (en) | 2024-01-23 |
Family
ID=88851079
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311379683.1A Active CN117132001B (en) | 2023-10-24 | 2023-10-24 | Multi-target wind control strategy optimization method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117132001B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117333290B (en) * | 2023-12-01 | 2024-03-26 | 杭银消费金融股份有限公司 | Integrated multi-scale wind control model construction method |
| CN117408805B (en) * | 2023-12-15 | 2024-03-22 | 杭银消费金融股份有限公司 | Credit wind control method and system based on stability modeling |
| CN117670149A (en) * | 2024-02-01 | 2024-03-08 | 杭银消费金融股份有限公司 | Passenger group quality scoring method |
| CN117853232B (en) * | 2024-03-07 | 2024-05-24 | 杭银消费金融股份有限公司 | Credit risk abnormity inspection attribution early warning method and system |
| CN118410319B (en) * | 2024-07-02 | 2024-09-03 | 杭银消费金融股份有限公司 | Method and system for generating and processing optimized wind control strategy |
Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108615188A (en) * | 2018-03-22 | 2018-10-02 | 深圳光华普惠科技有限公司 | A kind of system and method that net borrows risk management |
| CN110992173A (en) * | 2020-03-04 | 2020-04-10 | 杭州信雅达数码科技有限公司 | Credit risk assessment model generation method based on multi-instance learning |
| WO2021057130A1 (en) * | 2019-09-26 | 2021-04-01 | 支付宝(杭州)信息技术有限公司 | Intelligent risk control decision-making method and system, service processing method and system |
| CN113052676A (en) * | 2021-03-22 | 2021-06-29 | 广东贤能数字科技有限公司 | Intelligent wind control decision method, device, equipment and readable storage medium |
| WO2021168617A1 (en) * | 2020-02-24 | 2021-09-02 | 深圳市欢太科技有限公司 | Processing method and apparatus for service risk management, electronic device, and storage medium |
| CN113344700A (en) * | 2021-07-27 | 2021-09-03 | 上海华瑞银行股份有限公司 | Wind control model construction method and device based on multi-objective optimization and electronic equipment |
| CN113989019A (en) * | 2021-10-27 | 2022-01-28 | 平安银行股份有限公司 | Method, device, equipment and storage medium for identifying risks |
| CN114298829A (en) * | 2021-12-28 | 2022-04-08 | 上海数禾信息科技有限公司 | Data processing method and device for credit assessment |
| CN114841801A (en) * | 2022-07-04 | 2022-08-02 | 天津金城银行股份有限公司 | Credit wind control method and device based on user behavior characteristics |
| CN115062734A (en) * | 2022-08-19 | 2022-09-16 | 深圳前海环融联易信息科技服务有限公司 | Wind control modeling method, device, equipment and medium capable of outputting explanatory information |
| EP4105872A2 (en) * | 2021-11-10 | 2022-12-21 | Beijing Baidu Netcom Science And Technology Co. Ltd. | Data processing method and apparatus |
| CN115907968A (en) * | 2022-12-20 | 2023-04-04 | 杭银消费金融股份有限公司 | Wind control rejection inference method and device based on pedestrian credit |
| CN116739742A (en) * | 2023-06-02 | 2023-09-12 | 北京百度网讯科技有限公司 | Monitoring method, device, equipment and storage medium of credit wind control model |
| CN116805265A (en) * | 2023-06-28 | 2023-09-26 | 平安银行股份有限公司 | Wind control threshold selection method and device, computer equipment and readable storage medium |
| CN116823437A (en) * | 2023-06-14 | 2023-09-29 | 广东企企通科技有限公司 | Access method, device, equipment and medium based on configured wind control strategy |
| CN116843446A (en) * | 2023-05-22 | 2023-10-03 | 中山大学 | Method for evaluating and attributing infraction risk of small amount credit based on interpretable machine learning |
| CN116883154A (en) * | 2023-07-26 | 2023-10-13 | 深圳前海微众银行股份有限公司 | Credit risk identification method, credit risk identification device, electronic equipment and readable storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112418580A (en) * | 2019-08-22 | 2021-02-26 | 上海哔哩哔哩科技有限公司 | Risk control method, computer equipment and readable storage medium |
-
2023
- 2023-10-24 CN CN202311379683.1A patent/CN117132001B/en active Active
Patent Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108615188A (en) * | 2018-03-22 | 2018-10-02 | 深圳光华普惠科技有限公司 | A kind of system and method that net borrows risk management |
| WO2021057130A1 (en) * | 2019-09-26 | 2021-04-01 | 支付宝(杭州)信息技术有限公司 | Intelligent risk control decision-making method and system, service processing method and system |
| WO2021168617A1 (en) * | 2020-02-24 | 2021-09-02 | 深圳市欢太科技有限公司 | Processing method and apparatus for service risk management, electronic device, and storage medium |
| CN110992173A (en) * | 2020-03-04 | 2020-04-10 | 杭州信雅达数码科技有限公司 | Credit risk assessment model generation method based on multi-instance learning |
| CN113052676A (en) * | 2021-03-22 | 2021-06-29 | 广东贤能数字科技有限公司 | Intelligent wind control decision method, device, equipment and readable storage medium |
| CN113344700A (en) * | 2021-07-27 | 2021-09-03 | 上海华瑞银行股份有限公司 | Wind control model construction method and device based on multi-objective optimization and electronic equipment |
| CN113989019A (en) * | 2021-10-27 | 2022-01-28 | 平安银行股份有限公司 | Method, device, equipment and storage medium for identifying risks |
| EP4105872A2 (en) * | 2021-11-10 | 2022-12-21 | Beijing Baidu Netcom Science And Technology Co. Ltd. | Data processing method and apparatus |
| CN114298829A (en) * | 2021-12-28 | 2022-04-08 | 上海数禾信息科技有限公司 | Data processing method and device for credit assessment |
| CN114841801A (en) * | 2022-07-04 | 2022-08-02 | 天津金城银行股份有限公司 | Credit wind control method and device based on user behavior characteristics |
| CN115062734A (en) * | 2022-08-19 | 2022-09-16 | 深圳前海环融联易信息科技服务有限公司 | Wind control modeling method, device, equipment and medium capable of outputting explanatory information |
| CN115907968A (en) * | 2022-12-20 | 2023-04-04 | 杭银消费金融股份有限公司 | Wind control rejection inference method and device based on pedestrian credit |
| CN116843446A (en) * | 2023-05-22 | 2023-10-03 | 中山大学 | Method for evaluating and attributing infraction risk of small amount credit based on interpretable machine learning |
| CN116739742A (en) * | 2023-06-02 | 2023-09-12 | 北京百度网讯科技有限公司 | Monitoring method, device, equipment and storage medium of credit wind control model |
| CN116823437A (en) * | 2023-06-14 | 2023-09-29 | 广东企企通科技有限公司 | Access method, device, equipment and medium based on configured wind control strategy |
| CN116805265A (en) * | 2023-06-28 | 2023-09-26 | 平安银行股份有限公司 | Wind control threshold selection method and device, computer equipment and readable storage medium |
| CN116883154A (en) * | 2023-07-26 | 2023-10-13 | 深圳前海微众银行股份有限公司 | Credit risk identification method, credit risk identification device, electronic equipment and readable storage medium |
Non-Patent Citations (3)
| Title |
|---|
| "全生命周期"式互联网信贷风控策略;卢华玮;《中国农村金融》;第82-83页 * |
| 基于大数据的信用风控在化工行业的应用研究;梅新芝;;当代化工研究(第04期);全文 * |
| 基于大数据的信用风控在化工行业的应用研究;梅新芝;《当代化工研究》(第04期);第144-145页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117132001A (en) | 2023-11-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117132001B (en) | Multi-target wind control strategy optimization method and system | |
| Ogor | Student academic performance monitoring and evaluation using data mining techniques | |
| Lee et al. | Assessing the lifetime performance index of Rayleigh products based on the Bayesian estimation under progressive type II right censored samples | |
| Lee et al. | Assessing the lifetime performance index of products with the exponential distribution under progressively type II right censored samples | |
| CN108596255A (en) | Take into account the prediction of result grader of the context-aware study of fairness | |
| CN112836750A (en) | System resource allocation method, device and equipment | |
| John et al. | A modified control chart for monitoring non-normal characteristics | |
| CN114707644A (en) | Method and device for training graph neural network | |
| CN117333290B (en) | Integrated multi-scale wind control model construction method | |
| CN117078403B (en) | Wind control decision method and system based on rule combination optimization | |
| Gazeloglu et al. | Comparison of weighted least squares and robust estimation in structural equation modeling of ordinal categorical data with larger sample sizes | |
| CN107274043B (en) | Quality evaluation method and device of prediction model and electronic equipment | |
| Baucks et al. | Simulating Policy Changes in Prerequisite-Free Curricula: A Supervised Data-Driven Approach. | |
| CN116992602A (en) | Reliability fuzzy evaluation method based on failure state characterization, electronic equipment and storage medium | |
| CN116796140A (en) | Abnormal analysis method, device, equipment and storage medium based on artificial intelligence | |
| CN112308278A (en) | Method, device, equipment and medium for optimizing prediction model | |
| CN115936104A (en) | Method and apparatus for training machine learning models | |
| Sheng et al. | A comparative study of data mining techniques in predicting consumers' credit card risk in banks | |
| CN113240323A (en) | Level evaluation method and device based on machine learning and related equipment | |
| CN113887816A (en) | Intelligent prediction method, device and equipment for social reserve of household appliances | |
| Palomo et al. | Learning the tandem network Lindley recursion | |
| Chimitova et al. | Goodness-of-fit testing for the degradation models in reliability analysis | |
| CN113627585A (en) | Single well index prediction method and system based on long-time and short-time memory neural network | |
| CN118069930B (en) | Investment dynamic recommendation method and system based on matching analysis | |
| CN118378177B (en) | Multi-classification model prediction distribution adjustment method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |