CN117078403A - Wind control decision method and system based on rule combination optimization - Google Patents
Wind control decision method and system based on rule combination optimization Download PDFInfo
- Publication number
- CN117078403A CN117078403A CN202311344725.8A CN202311344725A CN117078403A CN 117078403 A CN117078403 A CN 117078403A CN 202311344725 A CN202311344725 A CN 202311344725A CN 117078403 A CN117078403 A CN 117078403A
- Authority
- CN
- China
- Prior art keywords
- rule set
- initial
- determining
- rule
- user
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 57
- 238000005457 optimization Methods 0.000 title claims abstract description 22
- 239000013598 vector Substances 0.000 claims abstract description 131
- 238000011156 evaluation Methods 0.000 claims abstract description 105
- 238000012545 processing Methods 0.000 claims abstract description 66
- 238000011217 control strategy Methods 0.000 claims abstract description 20
- 238000000605 extraction Methods 0.000 claims abstract description 8
- 238000012937 correction Methods 0.000 claims description 16
- 238000012216 screening Methods 0.000 claims description 11
- 238000012163 sequencing technique Methods 0.000 claims description 2
- 230000003542 behavioural effect Effects 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 4
- 239000000523 sample Substances 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 230000007246 mechanism Effects 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
- 230000004075 alteration Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000008520 organization Effects 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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The application provides a wind control decision method and a system based on rule combination optimization, which belong to the technical field of financial information and specifically comprise the following steps: the feature extraction is carried out according to the basic information and the behavior information of the user to obtain the feature vector of the user, the quantity limiting value of the feature vector of the initial rule set and the maximum value of the feature vector of the rule set of the combined rule set are taken as constraint conditions, the global rule set of the initial rule set and the global rule of the rule set of the combined rule set are generated at least by using a risk grid method, the initial risk and the normal user of the current credit application user are obtained based on the initial rule set and the initial wind control strategy, the evaluation risk of different rule sets of the normal user under the combined rule set is obtained based on the combined rule set and the wind control strategy of the normal user, and the approval is carried out by combining the initial risk, so that the processing efficiency and the accuracy of credit approval are improved.
Description
Technical Field
The application belongs to the technical field of financial information, and particularly relates to a wind control decision method and system based on rule combination optimization.
Background
In order to improve the processing efficiency of the credit application, the credit application organization often automatically checks the credit application information of the user through a pre-established wind control strategy and obtains the processing effect of the credit application, but at the same time, because the data volume of the basic information and the personal behavior information of the user is larger, how to extract key features based on the information of the user has great significance for improving the processing efficiency of the risk strategy.
In order to realize feature extraction of a user in a credit application processing process, in an application patent CN202310348690.9, namely a risk feature extraction method, a device, an electronic device and a storage medium, user probe data and user identification data for fraud identification and credit behavior features for credit admission credit are aggregated to obtain aggregated risk features for fraud identification and/or credit admission credit, so that the processing efficiency of credit approval business is improved, but the following problems exist:
even if the risk features are aggregated, because the feature dimensions of the user probe data, the user identification data and the credit behavior features are higher, higher requirements are put forward on the processing capacity of the running server of the rule policy and the complexity of the rule policy, and meanwhile, the processing efficiency of the credit application is also lowered.
Aiming at the technical problems, the application provides a wind control decision method and a system based on rule combination optimization.
Disclosure of Invention
In order to achieve the purpose of the application, the application adopts the following technical scheme:
according to one aspect of the application, a method for wind-controlled decision-making based on rule combination optimization is provided.
The wind control decision method based on rule combination optimization is characterized by comprising the following steps:
s1, carrying out feature extraction according to basic information and behavior information of a user to obtain feature vectors of the user, and determining the quantity limiting value of the feature vectors of an initial rule set and the maximum value of the feature vectors of a rule set of a combined rule set based on application data of a historical credit application user and historical operation data of a credit application processing server;
s2, taking the quantity limiting value of the feature vectors of the initial rule set and the maximum value of the feature vectors of the rule set of the combined rule set as constraint conditions, and generating a global rule set of the initial rule set and a global rule set of the combined rule set at least by using a risk grid method;
s3, determining initial comprehensive evaluation values of the global rule set of the initial rule set and comprehensive evaluation values of the global rule set of the combined rule set based on the rejection rate, the lifting degree and the number of feature vectors of the global rule set respectively, and obtaining the numbers of the initial rule set, the combined rule set and the rule set of the combined rule set based on the initial comprehensive evaluation values and the comprehensive evaluation values;
s4, obtaining initial risks and normal users of the current trust application user based on an initial rule set and an initial wind control strategy, obtaining evaluation risks of different rule sets of the normal users under the combined rule set based on the combined rule set and the wind control strategy of the normal users, and carrying out approval by combining the initial risks.
The application has the beneficial effects that:
1. the determination of the feature vectors of the rule sets is realized from the angles of the processing reliability of the credit application, the processing quantity of the historical credit application and the processing pressure of the credit application processing server by the number limiting values of the feature vectors of the initial rule sets and the determination of the maximum value of the feature vectors of the rule sets of the combined rule sets, so that the processing efficiency of the credit application is ensured, the processing pressure of the server is reduced, and the processing accuracy of the credit application is also ensured.
2. By determining the initial comprehensive evaluation value of the global rule set of the initial rule set and the comprehensive evaluation value of the global rule set of the combined rule set, not only is the influence of rejection rate and the like of the global rule set on the processing accuracy of the trust approval considered, but also the difference of the number of the feature vectors of different global rule sets is considered.
3. Through the evaluation of the normal user and the approval passing through the user, the trust realizes the advanced screening of the normal user, improves the trust approval efficiency, and simultaneously further improves the processing accuracy of trust application through the comprehensive evaluation risk of a plurality of rule sets.
Further technical solutions include, but are not limited to, the user's occupation, income, marital status, and social security payment information.
Further technical solutions include, but are not limited to, the number of times the user's credit applications are applied, the information of the peer credit, the information of the historical repayment and the overdue information.
The further technical scheme is that the method for determining the global rule set of the initial rule set comprises the following steps:
risk trellis method: and generating the set of the feature vectors and indexes corresponding to the set of the feature vectors based on the feature vectors, and generating a global rule set through intersecting the two-dimensional or multi-dimensional feature vectors.
The further technical scheme is that when the initial risk of the current credit application user is smaller than a preset risk value, the current credit application user is determined to be a normal user.
The further technical scheme is that the method for determining the approval passing the user comprises the following steps:
acquiring the evaluation risk of different rule sets of the normal user under the combined rule set, and when the evaluation risk of different rule sets of the normal user under the combined rule set is greater than a preset evaluation risk, acquiring the evaluation risk of different rule sets of the normal user under the combined rule set, wherein the rule sets are the same as the rule sets with the preset evaluation risk:
determining that the normal user does not belong to an approval passing user;
when the evaluation risk of the normal user under the different rule sets under the combined rule set does not exist the rule set which is larger than the preset evaluation risk:
carrying out the maximum value of the evaluation risks of the different rule sets of the normal user under the combined rule set and the number of the rule sets larger than a preset risk value according to the evaluation risks of the different rule sets of the normal user under the combined rule set, and carrying out the determination of the trust approval risk of the normal user by combining the sum of the evaluation risks of the different rule sets of the normal user under the combined rule set;
and determining approval passing users through the trust approval risks and the initial risks of the normal users.
In a second aspect, the present application provides a wind control decision system based on rule combination optimization, and the wind control decision method based on rule combination optimization specifically includes:
the system comprises a feature vector quantity determining module, a global rule set generating module, a rule set determining module and a credit approval module;
the feature vector quantity determining module is responsible for carrying out feature extraction according to the basic information and the behavior information of the user to obtain a feature vector of the user, and carrying out quantity limiting value of the feature vector of the initial rule set and determination of the maximum value of the feature vector of the rule set of the combined rule set based on application data of a historical credit application user and historical operation data of a credit application processing server;
the global rule set generation module is responsible for generating a global rule set of the initial rule set and a global rule set of the combined rule set by at least using a risk grid method by taking a quantity limiting value of the feature vectors of the initial rule set and a maximum value of the feature vectors of the rule set of the combined rule set as constraint conditions;
the rule set determining module is responsible for determining an initial comprehensive evaluation value of the global rule set of the initial rule set and a comprehensive evaluation value of the global rule set of the combined rule set based on the rejection rate, the lifting degree and the number of feature vectors of the global rule set respectively, and obtaining the number of rule sets of the initial rule set, the combined rule set and the combined rule set based on the initial comprehensive evaluation value and the comprehensive evaluation value;
the trust approval module is responsible for obtaining initial risks and normal users of the current trust application user based on an initial rule set and an initial wind control strategy, obtaining evaluation risks of different rule sets of the normal users under the combined rule set based on the combined rule set and the wind control strategy of the normal users, and carrying out approval by combining the initial risks.
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 application. The objectives and other advantages of the application 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 application more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
The above and other features and advantages of the present application 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 method of wind control decision optimization based on rule combinations;
FIG. 2 is a flow chart of a method of number limit determination of feature vectors of an initial rule set;
FIG. 3 is a flow chart of a method of determining a maximum value of feature vectors of a rule set of a combined rule set;
FIG. 4 is a flow chart of a method of approving a determination by a user;
FIG. 5 is a framework diagram of a wind-controlled decision system optimized based on rule combinations.
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 rejection rate refers to the proportion of the model processing results of the users which should not be rejected to the number of the users which should not be rejected in all the numbers of the users which should not be rejected, and the rejection rate in the application refers to the proportion of the processing results of the credit application which should be passed to the users which should not be passed in the default model to the proportion of the processing results of the users which should not be passed to the users which should be passed in all the credit application.
The improvement degree is a multiple of the prediction capability of the model on the bad sample compared with random selection, and in the method, the specific trust application processing result is a ratio of the number of failed users predicted by the wind control strategy to the number of failed users randomly selected predicted by the failed users when the wind control strategy is not adopted.
The wind control strategy and the initial wind control strategy are models for predicting the trust processing risk of the user by adopting the feature vector, and particularly can adopt common regression models based on BP neural network, linear regression model, CNN convolutional neural network and the like. Specifically, the specific steps of predicting the risk of the trusted processing by adopting the linear regression model are as follows:
obtaining initial risks and normal users of the current trust application users based on an initial rule set and an initial wind control strategy;
1. taking an initial rule set of a user as input data, and taking an initial risk of the user as output prediction data;
2. and carrying out scatter diagram analysis, observing the relation between input data and output predicted data, selecting a proper regression equation type according to the observed data characteristics, and estimating parameters in the regression equation through a least square method.
3. And analyzing and checking the obtained regression model, observing whether the residual diagram is abnormal or not, and when the residual diagram is not abnormal, taking an initial rule set of the current credit application user as an input set to determine the initial risk of the current credit application user.
The applicant finds that when the credit giving approval mechanism builds the default model, because the influence factors and data of credit giving default of the user are more, the artificial combination of the original feature vectors is adopted, so that the processing efficiency of the credit giving default model is possibly slower, and meanwhile, the real default risk can not be accurately reflected by the credit giving default model.
In order to solve the technical problems, the applicant adopts the following technical scheme:
the method comprises the steps of screening users with higher default risks through an initial rule set, wherein the initial default set only comprises a certain number of feature vectors, and then evaluating the default risks of the screened users with lower default risks according to the construction of a combined rule set, wherein the combined rule set comprises a plurality of rule sets, and one rule set comprises a plurality of groups of feature vectors, so that the evaluation of the default risks of the users with lower default risks can be accurately realized through the construction of the combined rule set.
Specifically, firstly, the number of feature vectors of an initial rule set and the number of feature vectors of a rule set of a combined rule set can be determined through application data, such as the application number, of a historical credit application user and historical operation data, such as load rate, of a credit application processing server, so that the processing efficiency is guaranteed, and meanwhile, the requirement of hardware conditions is met.
The combination of the feature vectors is realized through a risk grid method, so that the global rule set of the initial rule set and the global rule set of the combination rule set are determined, on the basis, the initial comprehensive evaluation value of the global rule set of the initial rule set and the comprehensive evaluation value of the global rule set of the combination rule set can be determined through factors such as rejection rate of a prediction model, and further the initial rule set and the combination rule set can be obtained, and the trust application of a user is processed based on the initial rule set and the combination rule set.
The following will describe in detail from both aspects of the method class embodiment and the system class embodiment.
To solve the above problem, according to one aspect of the present application, as shown in fig. 1, there is provided a wind control decision method based on rule combination optimization according to one aspect of the present application, which is characterized by specifically including:
s1, carrying out feature extraction according to basic information and behavior information of a user to obtain feature vectors of the user, and determining the quantity limiting value of the feature vectors of an initial rule set and the maximum value of the feature vectors of a rule set of a combined rule set based on application data of a historical credit application user and historical operation data of a credit application processing server;
the basic information of the user includes, but is not limited to, occupation, income, marital status and social security payment information of the user.
It should be noted that, the behavior information of the user includes, but is not limited to, the number of times of credit applications of the user, the credit information of the peer, the historical loan repayment information and the overdue information.
In one possible embodiment, as shown in fig. 2, the method for determining the number of feature vectors of the initial rule set in step S1 includes:
s11, determining the correlation coefficient of the feature vector through the correlation condition of the feature vector and the default state of the historical credit application user, and determining the maximum limiting number of the feature vector of the initial rule set by combining a preset correlation coefficient threshold;
s12, determining an average value of application quantity of the historical credit application user in unit time according to the application data of the historical credit application user, and determining whether the maximum limiting quantity can be adopted as a quantity limiting value or not according to the duty ratio of a time period which is larger than the average value of the application quantity of the historical credit application user in unit time in a preset time period, if so, entering a next step, and if not, entering a step S14;
s13, determining the average load rate of the credit application processing server according to the historical operation data of the credit application processing server, determining whether the maximum limiting quantity can be adopted as a quantity limiting value or not according to the duty ratio of a time period which is larger than the average load rate in a preset time period, if so, determining the quantity limiting value of the feature vector of the initial rule set through the maximum limiting quantity, and if not, entering step S14;
s14, determining the quantity correction quantity of the historical credit application users according to the average value and the peak value of the application quantity of the historical credit application users in unit time and the duty ratio of a time period which is larger than the average value of the application quantity of the historical credit application users in unit time in a preset time period;
s15, determining the quantity correction quantity of the credit application processing servers according to the average load rate and the peak load rate of the credit application processing servers and the duty ratio of a time period which is larger than the average load rate in a preset time period, and determining the quantity limit value of the feature vector of the initial rule set by combining the maximum limit quantity and the quantity correction quantity of the historical credit application users.
Further, the determining of the maximum limiting number of the feature vectors of the initial rule set in combination with the preset correlation coefficient threshold specifically includes:
and sequencing the correlation coefficients of the feature vectors from large to small to obtain the sequence of the feature vectors, and determining the maximum limiting number of the feature vectors of the initial rule set according to the sequence of the feature vectors by taking the fact that the sum of the correlation coefficients of the feature vectors is larger than the preset correlation coefficient threshold as a target.
In another possible embodiment, the method for determining the number of feature vectors of the initial rule set in the step S1 is as follows:
determining the correlation coefficient of the feature vector through the correlation condition of the feature vector and the default state of the historical credit application user, and determining the maximum limiting number of the feature vector of the initial rule set by combining a preset correlation coefficient threshold;
determining the peak value of the application quantity of the historical credit application users in unit time according to the application data of the historical credit application users, and determining the load rate peak value of the credit application processing server according to the historical operation data of the credit application processing server;
when the peak value of the application quantity of the historical credit application users in unit time and the peak value of the load rate of the credit application processing server are both in a preset range:
determining the quantity limiting value of the feature vectors of the initial rule set through the maximum limiting quantity;
when any one of the peak value of the application number of the historical credit application users in unit time and the peak value of the load rate of the credit application processing server is not in a preset range:
determining the quantity correction quantity of the historical credit application users according to the average value and the peak value of the application quantity of the historical credit application users in unit time and the duty ratio of a time period which is larger than the average value of the application quantity of the historical credit application users in unit time in a preset time period;
when the load rate peak value of the trust application processing server is not in a preset range:
determining the quantity correction quantity of the credit application processing servers according to the average load rate and the peak load rate of the credit application processing servers and the duty ratio of a time period which is larger than the average load rate in a preset time period, and determining the quantity limit value of the feature vector of the initial rule set by combining the maximum limit quantity and the quantity correction quantity of the historical credit application users;
when the load rate peak value of the trust application processing server is within a preset range:
and determining the quantity limiting value of the feature vectors of the initial rule set through the quantity correction quantity of the maximum limiting quantity and the historical credit application users.
In one possible embodiment, as shown in fig. 3, the method for determining the maximum value of the feature vector of the rule set of the combined rule set is as follows:
determining the correlation coefficient of the feature vector through the correlation condition of the feature vector and the default state of the historical credit application user, and determining the maximum limiting number of the feature vector of the rule set of the combination rule set by combining a preset correlation coefficient threshold;
determining the quantity correction quantity of the historical credit application users according to the average value and the peak value of the application quantity of the historical credit application users in unit time and the duty ratio of a time period which is larger than the average value of the application quantity of the historical credit application users in unit time in a preset time period;
determining the quantity correction quantity of the credit application processing servers according to the average load rate and the peak load rate of the credit application processing servers and the duty ratio of a time period which is larger than the average load rate in a preset time period, and determining the maximum quantity limit value of the characteristic vector of the rule set of the combination rule set by combining the maximum limit quantity of the characteristic vector of the rule set of the combination rule set and the quantity correction quantity of the history credit application users;
and determining a processing time limit value of each rule set of the combined rule set based on the credit application processing delay of the credit application user and the number of preset rule sets of the combined rule set, and determining the maximum value of the feature vector of the rule set of the combined rule set by combining the maximum limit value of the feature vector of the rule set of the combined rule set.
In this embodiment, the determination of the feature vector of the rule set is implemented from the angles of the processing reliability of the credit application, the processing number of the history credit application and the processing pressure of the credit application processing server by the number limit value of the feature vector of the initial rule set and the determination of the maximum value of the feature vector of the rule set of the combined rule set, so that the processing efficiency of the credit application is ensured, the processing pressure of the server is reduced, and the processing accuracy of the credit application is also ensured.
S2, taking the quantity limiting value of the feature vectors of the initial rule set and the maximum value of the feature vectors of the rule set of the combined rule set as constraint conditions, and generating a global rule set of the initial rule set and a global rule set of the combined rule set at least by using a risk grid method;
specifically, the method for determining the global rule set of the initial rule set is as follows:
risk trellis method: and generating the set of the feature vectors and indexes corresponding to the set of the feature vectors based on the feature vectors, and generating a global rule set through intersecting the two-dimensional or multi-dimensional feature vectors.
S3, determining initial comprehensive evaluation values of the global rule set of the initial rule set and comprehensive evaluation values of the global rule set of the combined rule set based on the rejection rate, the lifting degree and the number of feature vectors of the global rule set respectively, and obtaining the numbers of the initial rule set, the combined rule set and the rule set of the combined rule set based on the initial comprehensive evaluation values and the comprehensive evaluation values;
in one possible embodiment, the method for determining the initial comprehensive evaluation value of the global rule set of the initial rule set is as follows:
acquiring the number of the feature vectors of the global rule set of the initial rule set, and determining a set rejection rate and a set lifting degree based on the number of the feature vectors of the global rule set of the initial rule set;
when the global rule set of the initial rule set with the rejection rate smaller than the set rejection rate and the promotion rate larger than the preset promotion rate exists:
determining the initial rule set through the number of feature vectors of the global rule set of the initial rule set;
when there is no global rule set of the initial rule set with the reject rate smaller than the set reject rate and the promotion rate greater than the preset promotion rate:
screening the global rule set based on the number of the feature vectors of the global rule set of the initial rule set and a preset number range to obtain a screening rule set;
and determining an initial comprehensive evaluation value of the screening rule set according to the rejection rate, the lifting degree and the number of the feature vectors of the screening rule set, and determining the initial rule set based on the initial comprehensive evaluation value.
Further, the determining the combination rule set and the number of the combination rule sets based on the comprehensive evaluation value specifically includes:
obtaining comprehensive evaluation values of the global rule sets of the combination rule sets and the preset number of the combination rule sets,
when the comprehensive evaluation values of the global rule sets of the combination rule sets larger than the number of the preset combination rule sets are all larger than the preset comprehensive evaluation value:
determining the number of the initial combination rule sets through the number of the preset combination rule sets, and determining the initial combination rule sets from large to small based on the comprehensive evaluation values of the global rule sets of the combination rule sets;
judging whether the sum of the comprehensive evaluation values of the initial combination rule sets is larger than a preset evaluation value sum, if so, determining the combination rule sets through the initial combination rule sets, and if not, taking the preset evaluation value sum as a target, and determining the combination rule sets from large to small through the comprehensive evaluation values of the global rule sets of the combination rule sets;
when the total evaluation value unevenness of the global rule set of the combination rule set larger than the number of the preset combination rule sets is not larger than the preset total evaluation value:
and determining the number of the combination rule sets through the number of the preset combination rule sets, and determining the combination rule sets from large to small based on the comprehensive evaluation value of the global rule set of the combination rule sets.
In this embodiment, by determining the initial comprehensive evaluation value of the global rule set of the initial rule set and the comprehensive evaluation value of the global rule set of the combination rule set, not only the influence of the rejection rate or the like of the global rule set on the processing accuracy of the trust approval is considered, but also the difference in the number of feature vectors of different global rule sets is considered.
S4, obtaining initial risks and normal users of the current trust application user based on an initial rule set and an initial wind control strategy, obtaining evaluation risks of different rule sets of the normal users under the combined rule set based on the combined rule set and the wind control strategy of the normal users, and carrying out approval by combining the initial risks.
Specifically, when the initial risk of the current credit application user is smaller than a preset risk value, determining that the current credit application user is a normal user.
In one possible embodiment, as shown in fig. 4, the method for determining the approval in step S4 is as follows:
s41, acquiring the evaluation risk of different rule sets of the normal user under the combined rule set, determining whether a rule set with the evaluation risk larger than a preset evaluation risk exists according to the evaluation risk of different rule sets of the normal user under the combined rule set, if so, determining that the normal user does not belong to an approval passing user, and if not, entering the next step;
s42, determining the sum of the evaluation risks of the normal user under the combined rule set through the evaluation risks of different rule sets of the normal user under the combined rule set, determining whether the trust risk of the normal user cannot meet the requirements according to the sum of the evaluation risks of the normal user under the combined rule set, if so, entering the next step, and if not, entering the step S44;
s43, acquiring the initial risk of the normal user, judging whether the initial risk of the normal user is larger than a preset evaluation risk, if so, determining that the normal user does not belong to an approval passing user, and if not, entering the next step;
s44, carrying out the maximum value of the evaluation risks of the different rule sets of the normal user under the combined rule set and the number of the rule sets larger than a preset risk value according to the evaluation risks of the different rule sets of the normal user under the combined rule set, and carrying out the determination of the trust approval risk of the normal user by combining the sum of the evaluation risks of the different rule sets of the normal user under the combined rule set;
s45, determining approval passing users through the trust approval risk and the initial risk of the normal users.
It should be noted that, the preset evaluation risk is determined according to the maximum value of evaluation risks of different rule sets of the offending users under the combined rule set in the historical credit application users of the credit approval mechanism.
In another possible embodiment, the method for determining the approval by the user is as follows:
acquiring the evaluation risk of different rule sets of the normal user under the combined rule set, and when the evaluation risk of different rule sets of the normal user under the combined rule set is greater than a preset evaluation risk, acquiring the evaluation risk of different rule sets of the normal user under the combined rule set, wherein the rule sets are the same as the rule sets with the preset evaluation risk:
determining that the normal user does not belong to an approval passing user;
when the evaluation risk of the normal user under the different rule sets under the combined rule set does not exist the rule set which is larger than the preset evaluation risk:
carrying out the maximum value of the evaluation risks of the different rule sets of the normal user under the combined rule set and the number of the rule sets larger than a preset risk value according to the evaluation risks of the different rule sets of the normal user under the combined rule set, and carrying out the determination of the trust approval risk of the normal user by combining the sum of the evaluation risks of the different rule sets of the normal user under the combined rule set;
and determining approval passing users through the trust approval risks and the initial risks of the normal users.
In this embodiment, through the evaluation of the normal user and the approval passing through the user, the trust service realizes the advanced screening of the normal user, improves the efficiency of trust service approval, and simultaneously further improves the processing accuracy of the trust service application by integrating the evaluation risks of a plurality of rule sets.
On the other hand, as shown in fig. 5, the present application provides a wind control decision system based on rule combination optimization, and the wind control decision method based on rule combination optimization specifically includes:
the system comprises a feature vector quantity determining module, a global rule set generating module, a rule set determining module and a credit approval module;
the feature vector quantity determining module is responsible for carrying out feature extraction according to the basic information and the behavior information of the user to obtain a feature vector of the user, and carrying out quantity limiting value of the feature vector of the initial rule set and determination of the maximum value of the feature vector of the rule set of the combined rule set based on application data of a historical credit application user and historical operation data of a credit application processing server;
the global rule set generation module is responsible for generating a global rule set of the initial rule set and a global rule set of the combined rule set by at least using a risk grid method by taking a quantity limiting value of the feature vectors of the initial rule set and a maximum value of the feature vectors of the rule set of the combined rule set as constraint conditions;
the rule set determining module is responsible for determining an initial comprehensive evaluation value of the global rule set of the initial rule set and a comprehensive evaluation value of the global rule set of the combined rule set based on the rejection rate, the lifting degree and the number of feature vectors of the global rule set respectively, and obtaining the number of rule sets of the initial rule set, the combined rule set and the combined rule set based on the initial comprehensive evaluation value and the comprehensive evaluation value;
the trust approval module is responsible for obtaining initial risks and normal users of the current trust application user based on an initial rule set and an initial wind control strategy, obtaining evaluation risks of different rule sets of the normal users under the combined rule set based on the combined rule set and the wind control strategy of the normal users, and carrying out approval by combining the initial risks.
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 (10)
1. The wind control decision method based on rule combination optimization is characterized by comprising the following steps:
extracting features according to basic information and behavior information of a user to obtain feature vectors of the user, and determining the quantity limiting value of the feature vectors of an initial rule set and the maximum value of the feature vectors of a rule set of a combined rule set based on application data of a historical credit application user and historical operation data of a credit application processing server;
the method comprises the steps of taking a quantity limiting value of characteristic vectors of an initial rule set and a maximum value of characteristic vectors of a rule set of a combined rule set as constraint conditions, and generating a global rule set of the initial rule set and a global rule set of the combined rule set at least by using a risk grid method;
determining an initial comprehensive evaluation value of a global rule set of the initial rule set and a comprehensive evaluation value of a global rule set of a combined rule set based on the rejection rate, the lifting degree and the number of feature vectors of the global rule set respectively, and obtaining the numbers of the initial rule set, the combined rule set and the rule set of the combined rule set based on the initial comprehensive evaluation value and the comprehensive evaluation value;
obtaining initial risks and normal users of the current trusted application user based on an initial rule set and an initial wind control strategy, obtaining evaluation risks of different rule sets of the normal users under the combined rule set based on the combined rule set and the wind control strategy of the normal users, and carrying out approval by combining the initial risks to determine through the users.
2. The method for the pneumatic control decision optimization based on the rule combination as claimed in claim 1, wherein the basic information of the user includes but is not limited to occupation, income, marital status and social security payment information of the user.
3. The method of claim 1, wherein the user's behavioral information includes, but is not limited to, the number of applications the user applies to, peer credit information, historical repayment information, and overdue information.
4. The method for determining the wind control decision based on rule combination optimization according to claim 1, wherein the method for determining the number limit value of the feature vectors of the initial rule set is as follows:
s11, determining the correlation coefficient of the feature vector through the correlation condition of the feature vector and the default state of the historical credit application user, and determining the maximum limiting number of the feature vector of the initial rule set by combining a preset correlation coefficient threshold;
s12, determining an average value of application quantity of the historical credit application user in unit time according to the application data of the historical credit application user, and determining whether the maximum limiting quantity can be adopted as a quantity limiting value or not according to the duty ratio of a time period which is larger than the average value of the application quantity of the historical credit application user in unit time in a preset time period, if so, entering a next step, and if not, entering a step S14;
s13, determining the average load rate of the credit application processing server according to the historical operation data of the credit application processing server, determining whether the maximum limiting quantity can be adopted as a quantity limiting value or not according to the duty ratio of a time period which is larger than the average load rate in a preset time period, if so, determining the quantity limiting value of the feature vector of the initial rule set through the maximum limiting quantity, and if not, entering step S14;
s14, determining the quantity correction quantity of the historical credit application users according to the average value and the peak value of the application quantity of the historical credit application users in unit time and the duty ratio of a time period which is larger than the average value of the application quantity of the historical credit application users in unit time in a preset time period;
s15, determining the quantity correction quantity of the credit application processing servers according to the average load rate and the peak load rate of the credit application processing servers and the duty ratio of a time period which is larger than the average load rate in a preset time period, and determining the quantity limit value of the feature vector of the initial rule set by combining the maximum limit quantity and the quantity correction quantity of the historical credit application users.
5. The method for wind-controlled decision-making based on rule-based combinatorial optimization of claim 4, wherein determining the maximum defined number of feature vectors of the initial rule set in combination with a preset correlation coefficient threshold specifically comprises:
and sequencing the correlation coefficients of the feature vectors from large to small to obtain the sequence of the feature vectors, and taking the fact that the sum of the correlation coefficients of the feature vectors is larger than the preset correlation coefficient threshold as a target, and passing through the sequence of the feature vectors.
6. The method for determining a maximum value of feature vectors of a rule set of the combined rule set according to claim 1, wherein the method for determining the maximum value of feature vectors of the rule set comprises the following steps:
determining the correlation coefficient of the feature vector through the correlation condition of the feature vector and the default state of the historical credit application user, and determining the maximum limiting number of the feature vector of the rule set of the combination rule set by combining a preset correlation coefficient threshold;
determining the quantity correction quantity of the historical credit application users according to the average value and the peak value of the application quantity of the historical credit application users in unit time and the duty ratio of a time period which is larger than the average value of the application quantity of the historical credit application users in unit time in a preset time period;
determining the quantity correction quantity of the credit application processing servers according to the average load rate and the peak load rate of the credit application processing servers and the duty ratio of a time period which is larger than the average load rate in a preset time period, and determining the maximum quantity limit value of the characteristic vector of the rule set of the combination rule set by combining the maximum limit quantity of the characteristic vector of the rule set of the combination rule set and the quantity correction quantity of the history credit application users;
and determining a processing time limit value of each rule set of the combined rule set based on the credit application processing delay of the credit application user and the number of preset rule sets of the combined rule set, and determining the maximum value of the feature vector of the rule set of the combined rule set by combining the maximum limit value of the feature vector of the rule set of the combined rule set.
7. The method for determining a global rule set based on rule combination optimization according to claim 1, wherein the method for determining the global rule set of the initial rule set is as follows:
and generating the set of the feature vectors and indexes corresponding to the set of the feature vectors based on the feature vectors, and generating a global rule set through intersecting the two-dimensional or multi-dimensional feature vectors.
8. The method for determining an initial comprehensive evaluation value of a global rule set of the initial rule set according to claim 1, wherein the method for determining the initial comprehensive evaluation value of the global rule set based on rule combination optimization is as follows:
acquiring the number of the feature vectors of the global rule set of the initial rule set, and determining a set rejection rate and a set lifting degree based on the number of the feature vectors of the global rule set of the initial rule set;
when the global rule set of the initial rule set with the rejection rate smaller than the set rejection rate and the promotion rate larger than the preset promotion rate exists:
determining the initial rule set through the number of feature vectors of the global rule set of the initial rule set;
when there is no global rule set of the initial rule set with the reject rate smaller than the set reject rate and the promotion rate greater than the preset promotion rate:
screening the global rule set based on the number of the feature vectors of the global rule set of the initial rule set and a preset number range to obtain a screening rule set;
and determining an initial comprehensive evaluation value of the screening rule set according to the rejection rate, the lifting degree and the number of the feature vectors of the screening rule set, and determining the initial rule set based on the initial comprehensive evaluation value.
9. The method for determining the wind control decision based on rule combination optimization according to claim 1, wherein the method for determining the approval by the user is as follows:
s41, acquiring the evaluation risk of different rule sets of the normal user under the combined rule set, determining whether a rule set with the evaluation risk larger than a preset evaluation risk exists according to the evaluation risk of different rule sets of the normal user under the combined rule set, if so, determining that the normal user does not belong to an approval passing user, and if not, entering the next step;
s42, determining the sum of the evaluation risks of the normal user under the combined rule set through the evaluation risks of different rule sets of the normal user under the combined rule set, determining whether the trust risk of the normal user cannot meet the requirements according to the sum of the evaluation risks of the normal user under the combined rule set, if so, entering the next step, and if not, entering the step S44;
s43, acquiring the initial risk of the normal user, judging whether the initial risk of the normal user is larger than a preset evaluation risk, if so, determining that the normal user does not belong to an approval passing user, and if not, entering the next step;
s44, carrying out the maximum value of the evaluation risks of the different rule sets of the normal user under the combined rule set and the number of the rule sets larger than a preset risk value according to the evaluation risks of the different rule sets of the normal user under the combined rule set, and carrying out the determination of the trust approval risk of the normal user by combining the sum of the evaluation risks of the different rule sets of the normal user under the combined rule set;
s45, determining approval passing users through the trust approval risk and the initial risk of the normal users.
10. A rule combination optimization-based wind control decision system, which adopts the wind control decision method based on the rule combination optimization as claimed in any one of claims 1 to 9, and is characterized by comprising the following specific steps:
the system comprises a feature vector quantity determining module, a global rule set generating module, a rule set determining module and a credit approval module;
the feature vector quantity determining module is responsible for carrying out feature extraction according to the basic information and the behavior information of the user to obtain a feature vector of the user, and carrying out quantity limiting value of the feature vector of the initial rule set and determination of the maximum value of the feature vector of the rule set of the combined rule set based on application data of a historical credit application user and historical operation data of a credit application processing server;
the global rule set generation module is responsible for generating a global rule set of the initial rule set and a global rule set of the combined rule set by at least using a risk grid method by taking a quantity limiting value of the feature vectors of the initial rule set and a maximum value of the feature vectors of the rule set of the combined rule set as constraint conditions;
the rule set determining module is responsible for determining an initial comprehensive evaluation value of the global rule set of the initial rule set and a comprehensive evaluation value of the global rule set of the combined rule set based on the rejection rate, the lifting degree and the number of feature vectors of the global rule set respectively, and obtaining the number of rule sets of the initial rule set, the combined rule set and the combined rule set based on the initial comprehensive evaluation value and the comprehensive evaluation value;
the trust approval module is responsible for obtaining initial risks and normal users of the current trust application user based on an initial rule set and an initial wind control strategy, obtaining evaluation risks of different rule sets of the normal users under the combined rule set based on the combined rule set and the wind control strategy of the normal users, and carrying out approval by combining the initial risks.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311344725.8A CN117078403B (en) | 2023-10-18 | 2023-10-18 | Wind control decision method and system based on rule combination optimization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311344725.8A CN117078403B (en) | 2023-10-18 | 2023-10-18 | Wind control decision method and system based on rule combination optimization |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117078403A true CN117078403A (en) | 2023-11-17 |
| CN117078403B CN117078403B (en) | 2024-01-12 |
Family
ID=88704757
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311344725.8A Active CN117078403B (en) | 2023-10-18 | 2023-10-18 | Wind control decision method and system based on rule combination optimization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117078403B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117709906A (en) * | 2024-02-04 | 2024-03-15 | 杭银消费金融股份有限公司 | External data source query decision method and device |
| CN117709906B (en) * | 2024-02-04 | 2024-05-14 | 杭银消费金融股份有限公司 | External data source query decision method and device |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110827032A (en) * | 2019-09-26 | 2020-02-21 | 支付宝(杭州)信息技术有限公司 | Intelligent wind control decision method and system and service processing method and system |
| US20210019657A1 (en) * | 2018-03-23 | 2021-01-21 | Welab Information Technology (Shenzhen) Limited | Fraud detection and risk assessment method, system, device, and storage medium |
| CN113302638A (en) * | 2018-10-29 | 2021-08-24 | 强力交易投资组合2018有限公司 | Method and system for improving machines and systems for automatically performing distributed ledger and other transactions in spot and forward markets for energy, computing, storage, and other resources |
| CN114140259A (en) * | 2021-11-29 | 2022-03-04 | 中国平安财产保险股份有限公司 | Artificial intelligence-based wind control method, device, equipment and storage medium for underwriting |
| CN115186973A (en) * | 2022-05-27 | 2022-10-14 | 民生科技有限责任公司 | Credit line assessment method and cascade system based on multi-platform combined modeling |
| CN115482084A (en) * | 2022-09-14 | 2022-12-16 | 蚂蚁区块链科技(上海)有限公司 | Method and device for generating wind control rule set |
| CN115601133A (en) * | 2022-10-10 | 2023-01-13 | 亲家网络技术（北京）有限公司（Cn） | Financial credit wind control rule generation method, device, equipment and storage medium |
| CN115660822A (en) * | 2022-10-17 | 2023-01-31 | 浙江惠瀜网络科技有限公司 | Wind control strategy processing method and device for financial business, electronic equipment and storage medium |
| CN116011883A (en) * | 2023-01-31 | 2023-04-25 | 厦门国际银行股份有限公司 | Optimization method, system and storage medium of financial wind control decision rule |
| CN116109139A (en) * | 2023-01-20 | 2023-05-12 | 阿里云计算有限公司 | Wind control strategy generation method, decision method, server and storage medium |
| CN116151965A (en) * | 2023-04-04 | 2023-05-23 | 成都新希望金融信息有限公司 | Risk feature extraction method and device, electronic equipment and storage medium |
| CN116167629A (en) * | 2023-02-13 | 2023-05-26 | 重庆富民银行股份有限公司 | Distributed policy automatic generation method, system and storage medium |
| CN116797357A (en) * | 2023-08-24 | 2023-09-22 | 杭银消费金融股份有限公司 | Financial terminal-based credit authorization processing method and equipment |
-
2023
- 2023-10-18 CN CN202311344725.8A patent/CN117078403B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210019657A1 (en) * | 2018-03-23 | 2021-01-21 | Welab Information Technology (Shenzhen) Limited | Fraud detection and risk assessment method, system, device, and storage medium |
| CN113302638A (en) * | 2018-10-29 | 2021-08-24 | 强力交易投资组合2018有限公司 | Method and system for improving machines and systems for automatically performing distributed ledger and other transactions in spot and forward markets for energy, computing, storage, and other resources |
| CN110827032A (en) * | 2019-09-26 | 2020-02-21 | 支付宝(杭州)信息技术有限公司 | Intelligent wind control decision method and system and service processing method and system |
| 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 |
| CN114140259A (en) * | 2021-11-29 | 2022-03-04 | 中国平安财产保险股份有限公司 | Artificial intelligence-based wind control method, device, equipment and storage medium for underwriting |
| CN115186973A (en) * | 2022-05-27 | 2022-10-14 | 民生科技有限责任公司 | Credit line assessment method and cascade system based on multi-platform combined modeling |
| CN115482084A (en) * | 2022-09-14 | 2022-12-16 | 蚂蚁区块链科技(上海)有限公司 | Method and device for generating wind control rule set |
| CN115601133A (en) * | 2022-10-10 | 2023-01-13 | 亲家网络技术（北京）有限公司（Cn） | Financial credit wind control rule generation method, device, equipment and storage medium |
| CN115660822A (en) * | 2022-10-17 | 2023-01-31 | 浙江惠瀜网络科技有限公司 | Wind control strategy processing method and device for financial business, electronic equipment and storage medium |
| CN116109139A (en) * | 2023-01-20 | 2023-05-12 | 阿里云计算有限公司 | Wind control strategy generation method, decision method, server and storage medium |
| CN116011883A (en) * | 2023-01-31 | 2023-04-25 | 厦门国际银行股份有限公司 | Optimization method, system and storage medium of financial wind control decision rule |
| CN116167629A (en) * | 2023-02-13 | 2023-05-26 | 重庆富民银行股份有限公司 | Distributed policy automatic generation method, system and storage medium |
| CN116151965A (en) * | 2023-04-04 | 2023-05-23 | 成都新希望金融信息有限公司 | Risk feature extraction method and device, electronic equipment and storage medium |
| CN116797357A (en) * | 2023-08-24 | 2023-09-22 | 杭银消费金融股份有限公司 | Financial terminal-based credit authorization processing method and equipment |
Non-Patent Citations (3)
| Title |
|---|
| HECKMANN: "A critical review on supply chain risk - Definition, measure and modeling", 《OMEGA-INTERNATIONAL JOURNAL OF MANAGEMENT SCIENCE》, vol. 52, pages 119 - 132 * |
| 张鲁男;常宝岗;梅利;: "基于规则引擎及智能阈值的实时业务风控系统", 通信技术, no. 11, pages 146 - 150 * |
| 相瑞: "大数据背景下个人贷款授信风险控制探索与实践", 《金融纵横》, no. 515, pages 80 - 83 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117709906A (en) * | 2024-02-04 | 2024-03-15 | 杭银消费金融股份有限公司 | External data source query decision method and device |
| CN117709906B (en) * | 2024-02-04 | 2024-05-14 | 杭银消费金融股份有限公司 | External data source query decision method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117078403B (en) | 2024-01-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | Fairness under unawareness: Assessing disparity when protected class is unobserved | |
| US10943186B2 (en) | Machine learning model training method and device, and electronic device | |
| JP2018538587A (en) | Risk assessment method and system | |
| CN116342259A (en) | Automatic user credit rating method and device, electronic equipment and medium | |
| US11250368B1 (en) | Business prediction method and apparatus | |
| CN117132001B (en) | Multi-target wind control strategy optimization method and system | |
| Pavlenko et al. | Credit risk modeling using Bayesian networks | |
| CN112561685B (en) | Customer classification method and device | |
| Van Thiel et al. | Artificial intelligent credit risk prediction: An empirical study of analytical artificial intelligence tools for credit risk prediction in a digital era | |
| US20130218630A1 (en) | Method and system for analyzing and optimizing distribution of work from a plurality of queues | |
| CN109146667B (en) | Method for constructing external interface comprehensive application model based on quantitative statistics | |
| CN117078403B (en) | Wind control decision method and system based on rule combination optimization | |
| Mirtalaei et al. | A trust-based bio-inspired approach for credit lending decisions | |
| Shi et al. | A model for recognizing key factors and applications thereof to engineering | |
| CN114170000A (en) | Credit card user risk category identification method, device, computer equipment and medium | |
| WO2014179088A1 (en) | Apparatus and method for total loss prediction | |
| CN114219611A (en) | Loan amount calculation method and device, computer equipment and storage medium | |
| Lakshmi et al. | Machine learning approach for taxation analysis using classification techniques | |
| CN113822751A (en) | Online loan risk prediction method | |
| Huang et al. | Hidden technical debts for fair machine learning in financial services | |
| CN113822490B (en) | Asset collection method and device based on artificial intelligence and electronic equipment | |
| US20230222378A1 (en) | Method and system for evaluating fairness of machine learning model | |
| Babaei et al. | Which SME is worth an investment? An explainable machine learning approach | |
| CN116843343A (en) | Intelligent identification method and system for automobile financial risk | |
| Eponeshnikov et al. | Balancing Accuracy, Fairness and Privacy in Machine Learning through Adversarial Learning |
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 |