CN113191888A - Method and device for scoring by urging collection - Google Patents
Method and device for scoring by urging collection Download PDFInfo
- Publication number
- CN113191888A CN113191888A CN202110578935.8A CN202110578935A CN113191888A CN 113191888 A CN113191888 A CN 113191888A CN 202110578935 A CN202110578935 A CN 202110578935A CN 113191888 A CN113191888 A CN 113191888A
- Authority
- CN
- China
- Prior art keywords
- variable
- target
- collection
- scoring
- model
- 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.)
- Pending
Links
Images
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
Abstract
The application provides an income-promoting scoring method and device, wherein the method comprises the following steps: the method comprises the steps of obtaining user information of a target user, analyzing the user information, determining an overdue repayment day range to which overdue repayment days of the target user belong, determining an income promoting scoring model corresponding to the overdue repayment day range in each pre-constructed income promoting scoring model as a target income promoting scoring model, and inputting the user information into the target income promoting scoring model to obtain an income promoting scoring result of the target user. In the technical scheme, the collection urging scoring models corresponding to different overdue repayment day ranges are constructed in advance, the target collection urging scoring models are determined from the pre-constructed collection urging scoring models based on the overdue repayment day ranges to which the overdue repayment days of the target users belong, and the collection urging scoring is performed on the target users based on the target collection urging scoring models, so that the collection urging scoring accuracy is improved, an applicable collection urging strategy can be determined based on collection urging scoring results, and accurate collection urging is realized.
Description
Technical Field
The application relates to the field of internet financial industry, in particular to an income-promoting scoring method and device.
Background
In the post-loan management work, the overdue user needs to be subjected to credit urging and receiving scoring, and then an urging and receiving strategy is determined based on the result of the credit urging and receiving scoring so as to urge to receive loans for the overdue user.
The inventor finds that the existing income-promoting scoring scheme carries out the income-promoting scoring on all overdue users by adopting the same income-promoting scoring strategy in the research process. However, the overdue payment days are different in range, the risks of overdue users are different in performance, and if the same collection urging scoring strategies are adopted, the accuracy of collection urging scoring can be affected, so that the determined collection urging strategies are not applicable, and accurate collection urging cannot be realized.
Disclosure of Invention
The application provides an income promoting scoring method and device, and aims to solve the problems that in an existing income promoting scoring scheme, the same income promoting scoring strategy is adopted for all overdue users to carry out income promoting scoring, the accuracy of the income promoting scoring can be influenced, the determined income promoting strategy is not applicable, and accurate income promoting cannot be realized.
In order to achieve the above object, the present application provides the following technical solutions:
a method for revenue scoring comprising:
acquiring user information of a target user; the target user is a user to be urged to receive scores;
analyzing the user information, and determining the overdue repayment day range to which the overdue repayment days of the overdue user belong;
determining an acceptance-urging scoring model corresponding to the overdue repayment day range in each pre-constructed acceptance-urging scoring model as a target acceptance-urging scoring model;
and inputting the user information into the target collection and grading model to obtain a collection and grading result of the target user.
Optionally, the method for inputting the user information into the target collection scoring model to obtain the collection scoring result of the target user includes:
acquiring a plurality of variable attributes of the target collection scoring model;
extracting information corresponding to each variable attribute from the user information;
and inputting the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
Optionally, the method for obtaining the collection-promoting scoring result of the target user by inputting the information corresponding to each variable attribute into the target collection-promoting scoring model includes:
calculating a characteristic vector of information corresponding to each variable attribute;
and inputting the characteristic vector of the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
The method described above, optionally, the process of constructing each revenue-inducing scoring model, includes:
collecting sample data of a plurality of overdue users, wherein each sample data comprises a plurality of variables;
classifying the sample data according to overdue repayment days to obtain a plurality of sample sets;
performing variable screening on each sample data in the sample set according to a preset variable screening strategy corresponding to the sample set to obtain target sample data corresponding to each sample data, and forming each target sample data into a target sample set corresponding to the sample set;
selecting a plurality of target sample data from the target sample set to form a training data set aiming at each target sample set, selecting a plurality of target sample data from the target sample set to form a testing data set, training a pre-constructed logistic regression model according to each target sample data included in the training data set, testing the trained logistic regression model according to each target sample data included in the testing data set, and determining the tested logistic regression model as an initial catalytic recovery scoring model of the target sample set after the trained logistic regression model passes the testing;
and aiming at each initial collection grading model, collecting a cross-time verification sample set of the initial collection grading model, verifying the initial collection grading model according to each cross-time verification sample data included in the cross-time verification sample set, and determining the initial collection grading model as a collection grading model after the initial collection grading model passes the verification.
Optionally, the above method, where variable screening is performed on each sample data in the sample set according to a preset variable screening policy corresponding to the sample set, to obtain target sample data corresponding to each sample data, includes:
calculating the missing rate and the information value of each variable in each sample data in the sample set, and if the missing rate is smaller than a preset missing threshold value and the information value is within a preset range, determining the variable as a first variable;
for each first variable, judging whether the first variable meets a preset first business logic rule corresponding to the first variable, and if so, determining the first variable as a second variable;
performing box separation processing on each second variable to obtain a plurality of variable boxes;
for each variable box, if the proportion result of the variable box is smaller than a preset proportion threshold, determining each second variable in the variable box as a third variable; wherein the proportion result of the variable box is used for indicating the proportion between the number of the second variables included in the variable box and the sum of the number of all the second variables;
aiming at each third variable, calculating a multiple collinearity index VIF between the third variable and each other third variable, and removing other third variables of which the VIF is larger than a preset threshold, wherein the other variables are the third variables except the third variable in each third variable;
determining the third variable meeting a preset second business logic rule corresponding to the sample set in the rest third variables as a target variable;
and forming the target variables into a target sample set corresponding to the sample set.
An acceptance scoring apparatus comprising:
the device comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring user information of a target user; the target user is a user to be urged to receive scores;
the first determining unit is used for analyzing the user information and determining the overdue repayment day range to which the overdue repayment day of the overdue user belongs;
the second determining unit is used for determining the income promoting scoring model corresponding to the overdue repayment day range in each pre-constructed income promoting scoring model as a target income promoting scoring model;
and the input unit is used for inputting the user information into the target collection and grading model to obtain a collection and grading result of the target user.
Optionally, the above apparatus, where the input unit is configured to input the user information into the target collection scoring model to obtain a collection scoring result of the target user, includes that the input unit is specifically configured to:
acquiring a plurality of variable attributes of the target collection scoring model;
extracting information corresponding to each variable attribute from the user information;
and inputting the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
Optionally, the above apparatus, where the input unit is configured to input the information corresponding to each variable attribute into the target collection scoring model to obtain the collection scoring result of the target user, and the input unit is specifically configured to:
calculating a characteristic vector of information corresponding to each variable attribute;
and inputting the characteristic vector of the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
The above apparatus, optionally, further comprises:
the device comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring sample data of a plurality of overdue users, and each sample data comprises a plurality of variables;
the classification unit is used for classifying the sample data according to overdue repayment days to obtain a plurality of sample sets;
the screening unit is used for carrying out variable screening on each sample data in the sample set according to a preset variable screening strategy corresponding to the sample set to obtain target sample data corresponding to each sample data, and forming each target sample data into a target sample set corresponding to the sample set;
the training unit is used for selecting a plurality of target sample data from the target sample set to form a training data set aiming at each target sample set, selecting a plurality of target sample data from the target sample set to form a test data set, training a pre-constructed logistic regression model according to each target sample data included in the training data set, testing the trained logistic regression model correspondingly according to each target sample data included in the test data set, and determining the tested logistic regression model as an initial collection scoring model of the target sample set after the trained logistic regression model passes the test;
the verification unit is used for acquiring a cross-time verification sample set of the initial revenue-prompting scoring model aiming at each initial revenue-prompting scoring model, verifying the initial revenue-prompting scoring model according to each cross-time verification sample data included in the cross-time verification sample set, and determining the initial revenue-prompting scoring model as the revenue-prompting scoring model after the initial revenue-prompting scoring model passes the verification.
Optionally, the above apparatus, where the training unit is configured to perform variable filtering on each sample data in the sample set according to a preset variable filtering strategy corresponding to the sample set, to obtain target sample data corresponding to each sample data, and the training unit is specifically configured to:
calculating the missing rate and the information value of each variable in each sample data in the sample set, and if the missing rate is smaller than a preset missing threshold value and the information value is within a preset range, determining the variable as a first variable;
for each first variable, judging whether the first variable meets a preset first business logic rule corresponding to the first variable, and if so, determining the first variable as a second variable;
performing box separation processing on each second variable to obtain a plurality of variable boxes;
for each variable box, if the proportion result of the variable box is smaller than a preset proportion threshold, determining each second variable in the variable box as a third variable; wherein the proportion result of the variable box is used for indicating the proportion between the number of the second variables included in the variable box and the sum of the number of all the second variables;
aiming at each third variable, calculating a multiple collinearity index VIF between the third variable and each other third variable, and removing other third variables of which the VIF is larger than a preset threshold, wherein the other variables are the third variables except the third variable in each third variable;
determining the third variable meeting a preset second business logic rule corresponding to the sample set in the rest third variables as a target variable;
and forming the target variables into a target sample set corresponding to the sample set.
A storage medium, the storage medium includes stored instructions, and when the instructions are executed, the storage medium controls a device in which the storage medium is located to execute the catalytic yield scoring method.
An electronic device comprising a memory and one or more instructions, wherein the one or more instructions are stored in the memory and configured to be executed by one or more processors to perform the above-described catalytic scoring method.
Compared with the prior art, the method has the following advantages:
the application provides an income-promoting scoring method and device, wherein the method comprises the following steps: the method comprises the steps of obtaining user information of a target user, analyzing the user information, determining an overdue repayment day range to which overdue repayment days of the target user belong, determining an income promoting scoring model corresponding to the overdue repayment day range in each pre-constructed income promoting scoring model as a target income promoting scoring model, and inputting the user information into the target income promoting scoring model to obtain an income promoting scoring result of the target user. Obviously, in the technical scheme, the collection prompting scoring models corresponding to different overdue payment day ranges are constructed in advance, the target collection prompting scoring models are determined from the pre-constructed collection prompting scoring models based on the overdue payment day ranges to which the overdue payment days of the target user belong, and the collection prompting scoring is performed on the target user based on the target collection prompting scoring models, so that collection prompting scoring is performed by using different collection prompting models according to the different overdue payment day ranges, the collection prompting scoring accuracy is improved, an applicable collection prompting strategy can be determined based on collection prompting scoring results, and accurate collection prompting is realized.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a flow chart of a method of a revenue-driven scoring method provided herein;
FIG. 2 is a flow chart of another method of a revenue scoring method provided herein;
FIG. 3 is a flow chart of another method of an acceptance scoring method provided herein;
FIG. 4 is a flow chart of another method of a revenue scoring method provided herein;
FIG. 5 is a schematic structural diagram of an apparatus for urging collection and scoring according to the present application;
fig. 6 is a schematic structural diagram of an electronic device provided in the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
The application is operational with numerous general purpose or special purpose computing device environments or configurations. For example: personal computers, server computers, hand-held or portable devices, tablet-type devices, multi-processor apparatus, distributed computing environments that include any of the above devices or equipment, and the like.
The embodiment of the application provides a collection-urging scoring method, which can be applied to various system platforms, wherein an execution main body of the collection-urging scoring method can run on a computer terminal or a processor of various mobile devices, and a flow chart of the collection-urging scoring method is shown in fig. 1, and specifically comprises the following steps:
s101, obtaining user information of a target user.
In this embodiment, in response to the request for urging to receive the score, user information of a target user is obtained, where the target user is a user to be urged to receive the score, and the user information includes, but is not limited to, account information, statement information, transaction flow information, credit card repayment information, and consumption information.
In this embodiment, the collection prompting scoring request is a request sent by a bank worker when the bank worker needs to prompt collection scoring for a target user.
And S102, analyzing the user information, and determining the overdue payment day range to which the overdue payment days of the target user belong.
In this embodiment, the predetermined range of the overdue payment days is, optionally, 0 to 30 days and 31 to 60 days.
In this embodiment, the user information of the overdue user is analyzed, the overdue date of the overdue user included in the user information is acquired, the overdue repayment days of the overdue user are calculated based on the overdue date, and the overdue repayment day range to which the overdue repayment days of the target user belong is determined based on the preset overdue repayment day range.
S103, determining the income promoting scoring model corresponding to the overdue repayment day range in each pre-constructed income promoting scoring model as a target income promoting scoring model.
In this embodiment, the income promoting scoring models corresponding to different ranges of the number of days of expiration are pre-constructed, and the different income promoting scoring models correspond to different ranges of the number of days of expiration.
In this embodiment, the extent of the overdue payment days corresponding to the overdue payment days is determined, that is, whether the extent of the overdue payment days is 1 to 30 days or 31 to 60 days is determined, so as to determine the collection-urging scoring model corresponding to the overdue payment days, and the collection-urging scoring model corresponding to the overdue payment days is determined as the target model in each pre-constructed collection-urging scoring model.
In this embodiment, referring to fig. 2, a process of constructing each revenue-inducing scoring model includes the following steps:
s201, collecting sample data of a plurality of overdue users, wherein each sample data comprises a plurality of variable data.
The method includes the steps of collecting sample data of a plurality of overdue users, wherein the sample data are data generated by overdue user history, and each sample data comprises a plurality of variables.
S202, classifying the sample data according to overdue repayment days to obtain a plurality of sample sets.
In this embodiment, the overdue repayment days corresponding to each sample data are determined, each sample data is classified according to the overdue repayment days to obtain a plurality of sample sets, optionally, each sample data can be classified into two types to obtain two sample sets, which are respectively a first sample set and a second sample set, the overdue repayment days corresponding to the sample data included in the first sample set belong to a first overdue repayment day range, the overdue repayment days corresponding to the sample data included in the second sample set belong to a second overdue repayment day range, optionally, the first overdue repayment range is 1 to 30 days, and the second overdue repayment day range is 31 to 60 days; that is, different sample sets correspond to different ranges of days of overdue repayment.
S203, performing variable screening on each sample data in the sample set according to a preset variable screening strategy corresponding to the sample set to obtain target sample data corresponding to each sample data, and forming each target sample data into a target sample set corresponding to the sample set.
In this embodiment, a plurality of variable screening policies are preset, and different variable screening policies correspond to different ranges of overdue repayment days.
In this embodiment, for each sample set, a preset variable screening policy corresponding to the sample set is obtained, and specifically, the preset variable screening policy corresponding to the overdue payment day range is determined according to the overdue payment day range corresponding to the sample set.
In this embodiment, for each sample set, variable screening is performed on each sample data in the sample set according to a variable screening policy corresponding to the sample set, and a result obtained by variable screening in each sample data is determined as a target sample data of the sample data.
It should be noted that, each sample data in the same sample set is subjected to variable screening according to the same variable screening strategy, and sample data in different sample sets is subjected to variable screening according to different variable screening strategies.
Referring to fig. 3, the process of performing variable screening on each sample data in the sample set according to a preset variable screening policy corresponding to the sample set to obtain target sample data corresponding to each sample data specifically includes the following steps:
s301, calculating the loss rate and the information value of each variable in each sample data in the sample set, and if the loss rate is smaller than a preset loss threshold value and the information value is within a preset range, determining the variable as a first variable.
In this embodiment, for each variable in each sample data in a sample set, the missing rate and the information value of the variable are calculated, and please refer to the prior art for a specific way of calculating the missing rate and the information value.
In this embodiment, for each variable in each sample data in the sample set, it is determined whether the loss rate of the variable is smaller than a preset loss threshold, and whether the information value is within a preset range, optionally, the loss threshold may be 80%, and the preset range may be a range greater than 0.01 and smaller than 10.
In this embodiment, for each variable in each sample data in the sample set, if the loss rate of the variable is smaller than a preset loss threshold and the information value is within a preset range, the variable is determined as the first variable, otherwise, the variable is not determined as the first variable, that is, when the loss rate of the variable is not smaller than the preset loss threshold or the information value is not within the preset range, the variable is not determined as the first variable.
S302, judging whether the first variable accords with a preset first business logic rule corresponding to the first variable or not aiming at each first variable, and if so, determining the first variable as a second variable.
In this embodiment, first business logic rules corresponding to different variables are preset.
In this embodiment, for each first variable, a first business logic rule corresponding to the first variable is obtained, whether the first variable meets the first business logic rule corresponding to the first variable is determined, if the first variable meets the first business logic rule corresponding to the first variable, the first variable is determined as a second variable, and if the first variable does not meet the first business logic rule corresponding to the first variable, the first variable is not determined as the second variable.
And S303, performing box separation processing on each second variable to obtain a plurality of variable boxes.
And performing box separation on each second variable, specifically performing box separation on each second variable according to the data type of the second variable to obtain a plurality of variable boxes, wherein the data type includes but is not limited to a numerical type and a text type.
S304, for each variable box, if the proportion result of the variable box is smaller than a preset proportion threshold, determining each second variable in the variable box as a third variable.
In this embodiment, for each variable box, the proportion result of the variable box is calculated, specifically, the proportion result of the variable is obtained by dividing the number of the second variables included in the variable box by the sum of the numbers of all the second variables, that is, the proportion result of the variable box is used to indicate the proportion between the number of the second variables included in the variable box and the sum of the numbers of all the second variables.
In this embodiment, for each variable tank, it is determined whether the ratio result of the variable tank is smaller than a preset ratio threshold, if the ratio result of the variable tank is smaller than the preset ratio threshold, each second variable in the variable tank is determined as a third variable, and if the ratio result of the variable tank is not smaller than the preset ratio threshold, any one of the second variables in the variable tank is not determined as the third variable.
S305, calculating multiple collinearity indexes (VIF) between the third variable and each other third variable aiming at each third variable, and removing other third variables of which the VIF is larger than a preset threshold, wherein the other variables are the third variables except the third variable in each third variable.
In this embodiment, for each third variable, a multiple collinearity indicator VIF between the third variable and each other third variable is calculated, and a specific calculation manner of the multiple collinearity indicator VIF is please refer to the prior art, which is not described herein again. And the other third variables are the third variables except the third variable in each third variable.
In this embodiment, for each third variable, it is determined whether a multiple collinearity index VIF between the third variable and each of the other third variables is greater than a preset threshold, and each of the other variables having a VIF greater than the preset threshold is removed. Alternatively, the preset threshold may be 10.
In this embodiment, linear regression is used to calculate the coefficient of variance expansion (VIF) for the multiple collinearity in the control variables. The calculation of VIF is to calculate R through the regression method of each third variable to other third variables2，R2The larger the likelihood that variability accounting for a variable may be accounted for by other combinations of variables in the model.
Wherein, the larger the VIF is, the more likely that multiple collinearity occurs for the third variable and other third variables.
It should be noted that the third variable that has been eliminated is no longer used for the multiple collinearity index VIF of the calculation.
For each of the above third variables, calculating a multiple collinearity indicator VIF between the third variable and each of the other third variables, and performing a process of removing the other third variables whose VIF is greater than the preset threshold as follows:
and the third variables are A, B, C, D and E, and multiple collinearity VIF among AB, AC, AD and AE is respectively calculated aiming at the third variable A, wherein the VIF among AB is larger than a preset threshold value, the VIF among AE is larger than a preset threshold value, the third variables B and E are removed, then the multiple collinearity VIF among CDs is calculated aiming at the third variable C, and the third variable D is removed if the VIF among CDs is larger than the preset threshold value.
S306, determining the third variable meeting the preset second business logic rule corresponding to the sample set in the rest third variables as a target variable.
In this embodiment, it is determined whether the third variable satisfies a preset second business logic rule corresponding to the sample set to which the third variable belongs for each of the remaining third variables, and if the third variable satisfies the preset second business logic rule corresponding to the sample set to which the third variable belongs, the third variable is determined as the target variable.
In this embodiment, through the second business logic rule, the remaining third variables are further screened, and the variables that meet the business significance and have significant characteristics are screened out.
And S307, forming the target variables into a target sample set corresponding to the sample set.
In this embodiment, each target variable is combined into a target sample set corresponding to the sample set.
Optionally, for each sample data in the first sample set, according to a variable screening rule corresponding to the first sample set, screening 16 variables from each variable included in each sample data; and aiming at each sample data in the second sample set, screening 9 variables from all variables included in each second sample data according to a variable screening rule corresponding to the second sample set.
In this embodiment, each variable in each sample data in the sample set is subjected to deletion rate judgment, information value judgment, first business logic rule judgment, binning processing, multiple collinearity index VIF judgment and second business logic rule judgment, so that each variable is screened out to screen out a target variable.
S204, aiming at each target sample set, selecting a plurality of target sample data from the target sample set to form a training data set, selecting a plurality of target sample data from the target sample set to form a testing data set, training a pre-constructed logistic regression model according to each target sample data included in the training data set, testing the trained logistic regression model correspondingly according to each target sample data included in the testing data set, and determining the tested logistic regression model as an initial collection scoring model of the target sample set after the trained logistic regression model passes the testing.
In the embodiment, for each target sample set, a plurality of target sample data are selected from the target sample set to form a training data set, and a plurality of target sample data are selected from the target sample set to form a test data set; optionally, 70% of target sample data may be selected from the target sample set to make a training data set, and 30% of target sample data may be selected from the target sample set to form a test data set; it should be noted that the sum of the number of target sample data included in the training data set and the number of target sample data included in the test data set is equal to the number of target sample data included in the target sample set.
In this embodiment, for each target sample set, training a pre-constructed logistic regression model according to each target sample data in a training data set corresponding to the target sample set, specifically, a logistic regression method may be adopted to train the logistic regression model, and after training is completed, testing the trained logistic regression model according to each target sample data in a test data set corresponding to the target sample set, specifically, obtaining an ROC curve and a KS value of the trained logistic regression model, where the ROC curve: a Receiver Operating characteristics (Receiver Operating characteristics) curve is a useful visualization tool to compare two classification models. The ROC curve shows the trade-off between true normal (TPR) and false normal (FPR) for a given model. KS (Kolmogorov-Smirnov): the index measures the difference between the accumulated fractions of the good and bad samples, the KS value ranges from 0% to 100%, and the discrimination criteria are as follows:
KS value < 20%; a difference;
KS value of 20% -40%; generally;
KS value of 41-50%; good;
KS value of 51% -75%; is very good;
KS value > 75%; too high, a careful validation model is required.
In this embodiment, based on the ROC curve and the KS value, it is determined that all of the trained logistic regression models pass the test, and if the trained logistic regression models pass the test, the trained logistic regression models are determined as the initial collection scoring models of the target sample set.
S205, collecting a cross-time verification sample set of the initial revenue-prompting scoring model aiming at each initial revenue-prompting scoring model, verifying the initial revenue-prompting scoring model according to each cross-time verification sample data included in the cross-time verification sample set, and determining the initial revenue-prompting scoring model as the revenue-prompting scoring model after the initial revenue-prompting scoring model passes the verification.
In this embodiment, for each initial collection and grading model, a cross-time verification sample set of the initial collection and grading model is collected, the cross-time verification set is repayment behavior information of an overdue user within a presentation period, the presentation period is a time of pushing an observation point for N days, the observation point is a collection time point for collecting sample data of the overdue user, and it should be noted that N is different for different initial collection and grading models; for the initial collection-urging scoring model corresponding to the range of the overdue repayment days from 1 to 30 days, N can be set to be 90, and for the initial collection-urging scoring model corresponding to the range of the overdue repayment days from 31 to 60 days, N can be set to be 60.
The repayment behavior information comprises the performance of the overdue user in the performance period, and optionally, the performance of the overdue user in the performance period can be reflected as good or bad. In this embodiment, for different overdue payment days, a first expression classification table and a second expression classification table are set for defining the expression of the user in the expression period, the first expression classification table is used for defining the expression classification of the overdue user in the expression period corresponding to the overdue payment days ranging from 1 to 30 days, as shown in table 1, and the second expression classification table is used for defining the expression classification of the overdue user in the expression period corresponding to the overdue payment days ranging from 31 to 60 days, as shown in table 2.
TABLE 1 first expression Classification
TABLE 2 second expression Classification Table
And S104, inputting the user information into the target collection and grading model to obtain a collection and grading result of the target user.
In this embodiment, the user information is input into the target collection-urging scoring model to obtain the collection-urging scoring result of the target user output by the target collection-urging scoring model, so that the corresponding collection-urging strategy is determined based on the collection-urging scoring result, and accurate collection urging is realized. The collection-urging scoring result can be expressed as "good" or "bad", that is, the target collection-urging scoring model outputs good or bad collection-urging scoring result.
Referring to fig. 4, the process of inputting the user information into the target collection and scoring model to obtain the collection and scoring result of the target user includes:
s401, obtaining a plurality of variable attributes of the target collection scoring model.
In this embodiment, the variable attribute corresponding to each revenue-prompting scoring model is determined, that is, it is determined which variables need to be screened out as the input of the revenue-prompting scoring model.
In this embodiment, a plurality of variable attributes of the target collection scoring model are obtained, where the variable attributes are used to indicate types of variables to be input into the model.
S402, extracting information corresponding to each variable attribute from the user information.
Analyzing the user information, and determining the type of each variable included in the user information, so as to extract information corresponding to each variable attribute from the user information based on the type of each variable included in the user information, that is, for each variable attribute, extracting a variable with a type corresponding to the variable attribute from the user information.
And S403, inputting the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
In this embodiment, the information corresponding to each extracted variable attribute is input into the target collection scoring model, so as to obtain a collection scoring result of the target user.
Specifically, the process of inputting the information corresponding to each variable attribute into the target collection scoring model to obtain the collection scoring result of the target user specifically includes the following steps:
calculating a characteristic vector of information corresponding to each variable attribute;
and inputting the characteristic vector of the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
In this embodiment, the feature vector of the information corresponding to each variable attribute is obtained by performing the feature vector on the information corresponding to each variable attribute, and the feature vector of the information corresponding to each variable attribute is input into the target collection scoring model, so as to obtain the collection scoring result of the target user.
According to the income promoting scoring method provided by the embodiment of the application, the user information of the target user is obtained, the user information is analyzed, the range of the number of overdue repayment days of the target user is determined, the income promoting scoring model corresponding to the range of the overdue repayment days in each pre-constructed income promoting scoring model is determined to be the target income promoting scoring model, and the user information is input into the target income promoting scoring model to obtain the income promoting scoring result of the target user. By applying the collection urging scoring method provided by the embodiment of the application, collection urging scoring models corresponding to different overdue repayment day ranges are constructed in advance, the target collection urging scoring models are determined from the pre-constructed collection urging scoring models based on the overdue repayment day ranges to which the overdue repayment days of the target user belong, and collection urging scoring is performed on the target user based on the target collection urging scoring models, so that collection urging scoring is performed by using different collection urging models according to the different overdue repayment day ranges, the collection urging scoring accuracy is improved, an applicable collection urging strategy can be determined based on collection urging scoring results, and accurate collection urging is realized.
Corresponding to the method shown in fig. 1, an embodiment of the present application further provides a collection-urging scoring device, which is used for implementing the method shown in fig. 1, and a schematic structural diagram of the collection-urging scoring device is shown in fig. 5, and specifically includes:
an obtaining unit 501, configured to obtain user information of a target user; the target user is a user to be urged to receive scores;
a first determining unit 502, configured to analyze the user information, and determine a range of overdue payment days to which the overdue payment days of the overdue user belong;
a second determining unit 503, configured to determine, as a target revenue-prompting scoring model, a revenue-prompting scoring model corresponding to the overdue repayment day range in each pre-constructed revenue-prompting scoring model;
the input unit 504 is configured to input the user information into the target collection scoring model, so as to obtain a collection scoring result of the target user.
The utility model provides an urging to accept grading device, the number of days of overdue repayment scope that repayment days belongs to, confirm the target model of urging to accept from each receipts of establishing in advance grading model, urge to accept the model of grading based on the target, urge to accept the grade to the target user, realized urging to accept the number of days scope to different overdue repayment days, utilize different models of urging to accept to urge to accept to grade, improve the degree of accuracy that urges to accept to grade, thereby can be based on urging to accept the result of grading, determine the applicable strategy of urging to accept, realize accurate urging to accept.
In an embodiment of the application, based on the foregoing scheme, the input unit 504 is configured to input the user information into the target revenue-inducing rating model to obtain a revenue-inducing rating result of the target user, and includes the input unit 504 specifically configured to:
acquiring a plurality of variable attributes of the target collection scoring model;
extracting information corresponding to each variable attribute from the user information;
and inputting the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
In an embodiment of the application, based on the foregoing scheme, the input unit 504 is configured to input information corresponding to each variable attribute into the target collection scoring model, so as to obtain a collection scoring result of the target user, and includes that the input unit 504 is specifically configured to:
calculating a characteristic vector of information corresponding to each variable attribute;
and inputting the characteristic vector of the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
In an embodiment of the present application, based on the foregoing scheme, the method may further include:
the device comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring sample data of a plurality of overdue users, and each sample data comprises a plurality of variables;
the classification unit is used for classifying the sample data according to overdue repayment days to obtain a plurality of sample sets;
the screening unit is used for carrying out variable screening on each sample data in the sample set according to a preset variable screening strategy corresponding to the sample set to obtain target sample data corresponding to each sample data, and forming each target sample data into a target sample set corresponding to the sample set;
the training unit is used for selecting a plurality of target sample data from the target sample set to form a training data set aiming at each target sample set, selecting a plurality of target sample data from the target sample set to form a test data set, training a pre-constructed logistic regression model according to each target sample data included in the training data set, testing the trained logistic regression model correspondingly according to each target sample data included in the test data set, and determining the tested logistic regression model as an initial collection scoring model of the target sample set after the trained logistic regression model passes the test;
the verification unit is used for acquiring a cross-time verification sample set of the initial revenue-prompting scoring model aiming at each initial revenue-prompting scoring model, verifying the initial revenue-prompting scoring model according to each cross-time verification sample data included in the cross-time verification sample set, and determining the initial revenue-prompting scoring model as the revenue-prompting scoring model after the initial revenue-prompting scoring model passes the verification.
In an embodiment of the present application, based on the foregoing scheme, the training unit is configured to perform variable filtering on each sample data in the sample set according to a preset variable filtering strategy corresponding to the sample set, so as to obtain target sample data corresponding to each sample data, and the training unit is specifically configured to:
calculating the missing rate and the information value of each variable in each sample data in the sample set, and if the missing rate is smaller than a preset missing threshold value and the information value is within a preset range, determining the variable as a first variable;
for each first variable, judging whether the first variable meets a preset first business logic rule corresponding to the first variable, and if so, determining the first variable as a second variable;
performing box separation processing on each second variable to obtain a plurality of variable boxes;
for each variable box, if the proportion result of the variable box is smaller than a preset proportion threshold, determining each second variable in the variable box as a third variable; wherein the proportion result of the variable box is used for indicating the proportion between the number of the second variables included in the variable box and the sum of the number of all the second variables;
aiming at each third variable, calculating a multiple collinearity index VIF between the third variable and each other third variable, and removing other third variables of which the VIF is larger than a preset threshold, wherein the other variables are the third variables except the third variable in each third variable;
determining the third variable meeting a preset second business logic rule corresponding to the sample set in the rest third variables as a target variable;
and forming the target variables into a target sample set corresponding to the sample set.
An application embodiment further provides a storage medium, where the storage medium includes stored instructions, where the instructions, when executed, control a device in which the storage medium is located to perform the following operations:
acquiring user information of a target user; the target user is a user to be urged to receive scores;
analyzing the user information, and determining the overdue repayment day range to which the overdue repayment days of the overdue user belong;
determining an acceptance-urging scoring model corresponding to the overdue repayment day range in each pre-constructed acceptance-urging scoring model as a target acceptance-urging scoring model;
and inputting the user information into the target collection and grading model to obtain a collection and grading result of the target user.
The present embodiment further provides an electronic device, whose schematic structural diagram is shown in fig. 6, specifically including a memory 601, and one or more instructions 602, where the one or more instructions 602 are stored in the memory 601 and configured to be executed by one or more processors 603 to perform the following operations according to the one or more instructions 602:
acquiring user information of a target user; the target user is a user to be urged to receive scores;
analyzing the user information, and determining the overdue repayment day range to which the overdue repayment days of the overdue user belong;
determining an acceptance-urging scoring model corresponding to the overdue repayment day range in each pre-constructed acceptance-urging scoring model as a target acceptance-urging scoring model;
and inputting the user information into the target collection and grading model to obtain a collection and grading result of the target user.
It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. For the device-like embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
For convenience of description, the above devices are described as being divided into various units by function, and are described separately. Of course, the functionality of the units may be implemented in one or more software and/or hardware when implementing the present application.
From the above description of the embodiments, it is clear to those skilled in the art that the present application can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present application may be essentially or partially implemented in the form of a software product, which may be stored in a storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the embodiments or some parts of the embodiments of the present application.
The method and the device for hastening income and scoring provided by the application are described in detail above, a specific example is applied in the text to explain the principle and the implementation mode of the application, and the description of the above embodiment is only used for helping to understand the method and the core idea of the application; meanwhile, for a person skilled in the art, according to the idea of the present application, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present application.
Claims (10)
1. A collection-promoting scoring method, comprising:
acquiring user information of a target user; the target user is a user to be urged to receive scores;
analyzing the user information, and determining the overdue repayment day range to which the overdue repayment days of the overdue user belong;
determining an acceptance-urging scoring model corresponding to the overdue repayment day range in each pre-constructed acceptance-urging scoring model as a target acceptance-urging scoring model;
and inputting the user information into the target collection and grading model to obtain a collection and grading result of the target user.
2. The method according to claim 1, wherein the inputting the user information into the objective collection scoring model to obtain the collection scoring result of the objective user comprises:
acquiring a plurality of variable attributes of the target collection scoring model;
extracting information corresponding to each variable attribute from the user information;
and inputting the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
3. The method according to claim 2, wherein the inputting the information corresponding to each variable attribute into the target collection scoring model to obtain the collection scoring result of the target user comprises:
calculating a characteristic vector of information corresponding to each variable attribute;
and inputting the characteristic vector of the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
4. The method according to claim 1 or 3, wherein the construction process of each revenue scoring model comprises:
collecting sample data of a plurality of overdue users, wherein each sample data comprises a plurality of variables;
classifying the sample data according to overdue repayment days to obtain a plurality of sample sets;
performing variable screening on each sample data in the sample set according to a preset variable screening strategy corresponding to the sample set to obtain target sample data corresponding to each sample data, and forming each target sample data into a target sample set corresponding to the sample set;
selecting a plurality of target sample data from the target sample set to form a training data set aiming at each target sample set, selecting a plurality of target sample data from the target sample set to form a testing data set, training a pre-constructed logistic regression model according to each target sample data included in the training data set, testing the trained logistic regression model according to each target sample data included in the testing data set, and determining the tested logistic regression model as an initial catalytic recovery scoring model of the target sample set after the trained logistic regression model passes the testing;
and aiming at each initial collection grading model, collecting a cross-time verification sample set of the initial collection grading model, verifying the initial collection grading model according to each cross-time verification sample data included in the cross-time verification sample set, and determining the initial collection grading model as a collection grading model after the initial collection grading model passes the verification.
5. The method according to claim 4, wherein performing variable filtering on each sample data in the sample set according to a preset variable filtering policy corresponding to the sample set to obtain target sample data corresponding to each sample data includes:
calculating the missing rate and the information value of each variable in each sample data in the sample set, and if the missing rate is smaller than a preset missing threshold value and the information value is within a preset range, determining the variable as a first variable;
for each first variable, judging whether the first variable meets a preset first business logic rule corresponding to the first variable, and if so, determining the first variable as a second variable;
performing box separation processing on each second variable to obtain a plurality of variable boxes;
for each variable box, if the proportion result of the variable box is smaller than a preset proportion threshold, determining each second variable in the variable box as a third variable; wherein the proportion result of the variable box is used for indicating the proportion between the number of the second variables included in the variable box and the sum of the number of all the second variables;
aiming at each third variable, calculating a multiple collinearity index VIF between the third variable and each other third variable, and removing other third variables of which the VIF is larger than a preset threshold, wherein the other variables are the third variables except the third variable in each third variable;
determining the third variable meeting a preset second business logic rule corresponding to the sample set in the rest third variables as a target variable;
and forming the target variables into a target sample set corresponding to the sample set.
6. A collection and scoring device, comprising:
the device comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring user information of a target user; the target user is a user to be urged to receive scores;
the first determining unit is used for analyzing the user information and determining the overdue repayment day range to which the overdue repayment day of the overdue user belongs;
the second determining unit is used for determining the income promoting scoring model corresponding to the overdue repayment day range in each pre-constructed income promoting scoring model as a target income promoting scoring model;
and the input unit is used for inputting the user information into the target collection and grading model to obtain a collection and grading result of the target user.
7. The apparatus according to claim 6, wherein the input unit is configured to input the user information into the objective collection scoring model to obtain a collection scoring result of the objective user, and the input unit is specifically configured to:
acquiring a plurality of variable attributes of the target collection scoring model;
extracting information corresponding to each variable attribute from the user information;
and inputting the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
8. The apparatus according to claim 7, wherein the input unit is configured to input information corresponding to each variable attribute into the target collection scoring model to obtain a collection scoring result of the target user, and the input unit is specifically configured to:
calculating a characteristic vector of information corresponding to each variable attribute;
and inputting the characteristic vector of the information corresponding to each variable attribute into the target collection and grading model to obtain a collection and grading result of the target user.
9. The apparatus of claim 6 or 8, further comprising:
the device comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring sample data of a plurality of overdue users, and each sample data comprises a plurality of variables;
the classification unit is used for classifying the sample data according to overdue repayment days to obtain a plurality of sample sets;
the screening unit is used for carrying out variable screening on each sample data in the sample set according to a preset variable screening strategy corresponding to the sample set to obtain target sample data corresponding to each sample data, and forming each target sample data into a target sample set corresponding to the sample set;
the training unit is used for selecting a plurality of target sample data from the target sample set to form a training data set aiming at each target sample set, selecting a plurality of target sample data from the target sample set to form a test data set, training a pre-constructed logistic regression model according to each target sample data included in the training data set, testing the trained logistic regression model correspondingly according to each target sample data included in the test data set, and determining the tested logistic regression model as an initial collection scoring model of the target sample set after the trained logistic regression model passes the test;
the verification unit is used for acquiring a cross-time verification sample set of the initial revenue-prompting scoring model aiming at each initial revenue-prompting scoring model, verifying the initial revenue-prompting scoring model according to each cross-time verification sample data included in the cross-time verification sample set, and determining the initial revenue-prompting scoring model as the revenue-prompting scoring model after the initial revenue-prompting scoring model passes the verification.
10. The apparatus according to claim 9, wherein the training unit is configured to perform variable filtering on each sample data in the sample set according to a preset variable filtering policy corresponding to the sample set to obtain target sample data corresponding to each sample data, and the training unit is specifically configured to:
calculating the missing rate and the information value of each variable in each sample data in the sample set, and if the missing rate is smaller than a preset missing threshold value and the information value is within a preset range, determining the variable as a first variable;
for each first variable, judging whether the first variable meets a preset first business logic rule corresponding to the first variable, and if so, determining the first variable as a second variable;
performing box separation processing on each second variable to obtain a plurality of variable boxes;
for each variable box, if the proportion result of the variable box is smaller than a preset proportion threshold, determining each second variable in the variable box as a third variable; wherein the proportion result of the variable box is used for indicating the proportion between the number of the second variables included in the variable box and the sum of the number of all the second variables;
aiming at each third variable, calculating a multiple collinearity index VIF between the third variable and each other third variable, and removing other third variables of which the VIF is larger than a preset threshold, wherein the other variables are the third variables except the third variable in each third variable;
determining the third variable meeting a preset second business logic rule corresponding to the sample set in the rest third variables as a target variable;
and forming the target variables into a target sample set corresponding to the sample set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110578935.8A CN113191888A (en) | 2021-05-26 | 2021-05-26 | Method and device for scoring by urging collection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110578935.8A CN113191888A (en) | 2021-05-26 | 2021-05-26 | Method and device for scoring by urging collection |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113191888A true CN113191888A (en) | 2021-07-30 |
Family
ID=76985288
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110578935.8A Pending CN113191888A (en) | 2021-05-26 | 2021-05-26 | Method and device for scoring by urging collection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113191888A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116862668A (en) * | 2023-09-05 | 2023-10-10 | 杭州度言软件有限公司 | Intelligent collecting accelerating method for improving collecting accelerating efficiency |
-
2021
- 2021-05-26 CN CN202110578935.8A patent/CN113191888A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116862668A (en) * | 2023-09-05 | 2023-10-10 | 杭州度言软件有限公司 | Intelligent collecting accelerating method for improving collecting accelerating efficiency |
| CN116862668B (en) * | 2023-09-05 | 2023-11-24 | 杭州度言软件有限公司 | Intelligent collecting accelerating method for improving collecting accelerating efficiency |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108108902B (en) | Risk event warning method and device | |
| US20170004128A1 (en) | Device and method for analyzing reputation for objects by data mining | |
| CN110246031A (en) | Appraisal procedure, system, equipment and the storage medium of business standing | |
| CN111507831A (en) | Credit risk automatic assessment method and device | |
| WO2010037030A1 (en) | Evaluating loan access using online business transaction data | |
| JP5603468B1 (en) | Document sorting system, document sorting method, and document sorting program | |
| CN111709826A (en) | Target information determination method and device | |
| US20200065867A1 (en) | Patent valuation system | |
| CN111557011A (en) | Enterprise yield-breaking prediction system and operation method thereof | |
| CN111738819A (en) | Method, device and equipment for screening characterization data | |
| Svabova et al. | Prediction model of firms financial distress | |
| CN112598294A (en) | Method, device, machine readable medium and equipment for establishing scoring card model on line | |
| CN111583012B (en) | Method for evaluating default risk of credit, debt and debt main body by fusing text information | |
| CN111861521A (en) | Data processing method and device, computer readable medium and electronic equipment | |
| JP5986687B2 (en) | Data separation system, data separation method, program for data separation, and recording medium for the program | |
| CN112163944A (en) | Loan qualification scoring method and device for customer, computer equipment and storage medium | |
| JP5622969B1 (en) | Document analysis system, document analysis method, and document analysis program | |
| CN111275338A (en) | Method, device, equipment and storage medium for judging enterprise fraud behaviors | |
| CN112990989B (en) | Value prediction model input data generation method, device, equipment and medium | |
| CN113191888A (en) | Method and device for scoring by urging collection | |
| CN113159637A (en) | Malicious user determination method and device, storage medium and electronic device | |
| Yuan | Research on credit risk assessment of P2P network platform: based on the logistic regression model of evidence weight | |
| KR102336462B1 (en) | Apparatus and method of credit rating | |
| CN114493851A (en) | Risk processing method and device | |
| Lee et al. | The firm life cycle forecasting model using machine learning based on news articles |
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 |