CN109191282A - Methods of marking and system are monitored in a kind of loan of Behavior-based control model - Google Patents
Methods of marking and system are monitored in a kind of loan of Behavior-based control model Download PDFInfo
- Publication number
- CN109191282A CN109191282A CN201810966371.3A CN201810966371A CN109191282A CN 109191282 A CN109191282 A CN 109191282A CN 201810966371 A CN201810966371 A CN 201810966371A CN 109191282 A CN109191282 A CN 109191282A
- Authority
- CN
- China
- Prior art keywords
- account
- data
- variable
- prediction
- obtains
- 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
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 present invention, which discloses, monitors methods of marking and system in a kind of loan of Behavior-based control model, the described method comprises the following steps: obtaining the account initial data in the observation period；Feature extraction is carried out to the account initial data and obtains predictive variable；It is modeled according to predictive variable using machine learning algorithm, obtains the prediction model of response variable and predictive variable；According to the prediction model to there is the account refunded and showed to predict, the Default Probability in account time span of forecast is obtained；According to the Default Probability in the account time span of forecast, the behavior scoring of account is calculated.The present invention can increase data change speed, raising applicability and generalization, carry out accurately behavior scoring to account.
Description
Technical field
The present invention relates to credit financing field, more particularly to monitored in a kind of loan of Behavior-based control model methods of marking with
And system.
Background technique
Credit rating is also known as " borrowing letter grading " or credit assessment, is important content and the basis for establishing social credit system,
According to common definition, credit rating is credit rating service organization with third-party objective, just position, according to commenting for specification
Assessment system fulfils stringent appraisal procedure, to enterprise, financial institution, bond issuer with the appraisal procedure of science
Credit record, the quality of enterprise, managerial ability, management level, external environment, financial feelings of main body are participated in markets such as social organizations
Condition, development prospect etc. are fully understanded, after surveying and studying, analyzing and researching, with regard to it in the following energy met commitment for a period of time
Power, in fact it could happen that the overall merit done of various risks, and its superiority and inferiority is identified with certain meeting and is published in the public
A kind of economic activity, credit rating by loan application obligatio personalis repay risk evaluate, in order to gold such as banks
Melt mechanism and examination & approval credit is carried out to loan application people.
Traditional credit rating method is mostly based on expert's rule or scorecard model, i.e., formulates previously according to expertise
A set of code of points applies this set rule and carries out credit scoring, however, this credit rating further according to the real data of user
Mode is the scoring for having had experience to carry out based on history, and scoring has certain hysteresis quality, cannot react under the new situation new
User situation, and the specified and modification of its code of points requires the period one for being expounded through peer review by stringent, formulating and modifying
As it is long, data change speed is slow.
Summary of the invention
The purpose of the present invention is to provide one kind can increase data change speed, improves applicability and generalization, right
Account carries out monitoring methods of marking and system in the accurately loan of the Behavior-based control model of behavior scoring.
In order to achieve the above objectives, first aspect present invention proposes to monitor methods of marking in the loan of Behavior-based control model a kind of,
The following steps are included:
Obtain the account initial data in the observation period；
Feature extraction is carried out to the account initial data and obtains predictive variable；
It is modeled according to the predictive variable using machine learning algorithm, obtains the prediction of response variable and predictive variable
Model；
The promise breaking obtained in account time span of forecast is general to be predicted to the account for having performance of refunding according to the prediction model
Rate；
According to the Default Probability in the account time span of forecast, the behavior scoring of account is calculated.
Preferably, the account initial data includes account credit rating data and the practical refund situation data of account.
Preferably, described the step of obtaining predictive variable to account initial data progress feature extraction, includes:
Account initial credit ratings data, credit line behaviour in service data, close is obtained according to the account initial data
Phase refund behavioral data and the in the recent period variable data of this four dimensions data of overdue behavioral data；
Data prediction is carried out to the variable data and obtains the predictive variable.
Preferably, described the step of obtaining the predictive variable to variable data progress data prediction, includes:
Missing values processing and outlier processing are carried out to the variable data；
Calculate the emphasis index value of the variable data by missing values processing and outlier processing；
The emphasis index value of the variable data is compared with preset threshold, according to comparison result to the variable number
According to progress preliminary screening；
Variable data after preliminary screening is grouped processing；
Variable data after packet transaction is subjected to the conversion of WOE evidence weight；
Correlation analysis will be carried out by the variable data of WOE evidence weight conversion, demonstrate,proved based on the analysis results by WOE
Postsearch screening is carried out according to the variable data of weight conversion, obtains predictive variable.
Preferably, the title of the variable data is as shown in the table:
Preferably, described that the account for having performance of refunding is predicted in real time according to the prediction model, it is pre- to obtain account
The step of Default Probability in the survey phase includes:
Prediction index is filtered out from the predictive variable；
The prediction index is brought into and obtains the Default Probability in account time span of forecast in the prediction model；
Wherein, the prediction index includes:
Total overdue issue,
The maximum overdue number of days of history,
The last time refunds away from modern duration
Amount can be used
History it is the last it is overdue to term day duration _ maximum,
Total loaning bill capital account for accrediting amount ratio,
Acute conjunctivitis point 1.1,
Residue should go back capital/total loaning bill capital and
Current rainbow grading.
Preferably, further include the steps that having and prediction model is assessed by appraisal procedure, wherein the appraisal procedure
It include: one of Lorentz curve, ROC curve and KS curve or a variety of.
Preferably, the machine learning algorithm includes that logistic regression algorithm, random forests algorithm or gradient promote decision tree
One of algorithm is a variety of.
Second aspect of the present invention proposes to monitor points-scoring system in the loan of Behavior-based control model a kind of, comprising:
Module is obtained, for obtaining the account initial data in the observation period；
Characteristic extracting module carries out feature extraction to the account initial data and obtains predictive variable；
Modeling module is modeled according to the predictive variable, obtains the prediction model of response variable and predictive variable；
Prediction module obtains in account time span of forecast according to the prediction model to there is the account refunded and showed to predict
Default Probability；
Grading module calculates the behavior scoring of account according to the Default Probability in the account future time section.
Preferably, further includes:
Evaluation module, for assessing the prediction module.
Third aspect present invention proposes a kind of computer equipment, including memory, processor and storage are on a memory simultaneously
The computer program that can be run on a processor, the processor realize the method when executing described program.
Fourth aspect present invention proposes a kind of computer readable storage medium, stores in the computer readable storage medium
There is instruction, when the computer readable storage medium is run on computers, so that the computer executes the method.
Beneficial effects of the present invention are as follows:
The present invention is directed to the hysteresis quality of current existing credit rating mode, cannot react new user situation under the new situation,
And the specified and modification of its code of points requires to be expounded through peer review by stringent, the period formulated and modified is generally long,
The slow problem of data change speed has been formulated in a kind of loan of Behavior-based control model and has monitored methods of marking, calculated by machine learning
Method is modeled, and can be updated according to the iteration of data and be modeled again.Data change speed is increased, applicability is improved
And generalization, on the other hand, the present invention by account credit rating data and the practical refund situation data of account as
Account initial data can be improved the acquisition speed of data, improve work efficiency, and can carry out accurately behavior to account and comment
Point, be conducive to company borrow in policy management work, effectively reduce the credit risk of financial institution.
Detailed description of the invention
Specific embodiments of the present invention will be described in further detail with reference to the accompanying drawing.
Fig. 1 shows the stream that methods of marking is monitored in a kind of loan of Behavior-based control model described in one embodiment of the present of invention
Journey schematic diagram；
Fig. 2 shows being predicted account initial data progress feature extraction described in one embodiment of the present of invention
The flow diagram of variable；
Fig. 3 shows and obtains the prediction change to variable data progress data prediction described in one embodiment of the present of invention
The flow diagram of amount；
Fig. 4 shows pre- to there is the account refunded and showed to carry out according to prediction model described in one embodiment of the present of invention
It surveys, obtains the flow diagram of the Default Probability in account time span of forecast；
Fig. 5 shows the stream that points-scoring system is monitored in a kind of loan of Behavior-based control model described in one embodiment of the present of invention
Journey schematic diagram；
Fig. 6 shows a kind of structural schematic diagram of computer equipment described in one embodiment of the present of invention.
Specific embodiment
In order to illustrate more clearly of the present invention, the present invention is done further below with reference to preferred embodiments and drawings
It is bright.Similar component is indicated in attached drawing with identical appended drawing reference.It will be appreciated by those skilled in the art that institute is specific below
The content of description is illustrative and be not restrictive, and should not be limited the scope of the invention with this.
Credit rating is also known as " borrowing letter grading " or credit assessment, is important content and the basis for establishing social credit system,
According to common definition, credit rating is credit rating service organization with third-party objective, just position, according to commenting for specification
Assessment system, with science appraisal procedure, fulfil stringent appraisal procedure, to enterprise, financial institution, bond issuer and
The markets such as social organization participate in the credit record of main body, the quality of enterprise, managerial ability, management level, external environment, financial feelings
Condition, development prospect etc. are fully understanded, after surveying and studying, analyzing and researching, with regard to it in the following energy met commitment for a period of time
Power, in fact it could happen that the overall merit done of various risks, and its superiority and inferiority is identified with certain meeting and is published in the public
A kind of economic activity, credit rating by loan application obligatio personalis repay risk evaluate, in order to gold such as banks
Melt mechanism and examination & approval credit is carried out to loan application people.
Methods of marking, such as Fig. 1 are monitored in a kind of loan for Behavior-based control model that Fig. 1 proposes for one embodiment of the present of invention
It is shown, it the described method comprises the following steps:
S100, the account initial data in the acquisition observation period.
The observation period should be understood as obtaining the time range of account initial data herein, such as we will select the observation period
It is selected as over two months, that is, selection user goes over the account initial data in two months as sample.In the present embodiment
In specific implementation, account initial data mainly includes the account credit rating data and the practical refund situation number of account of user
According to, the account initial data of acquisition can be it is extraneous input in real time, be also possible to search in preset database and carry out
It arrives, database mentioned here is the database for storing account initial data.
By the way that number can be reduced using account credit rating data and the practical refund situation data of account as sample data
According to acquisition time, working efficiency is improved.
It further, can also be to account if using the account of user in same transaction data in the industry as sample data
Do the prediction of behavior scoring model.
S200, predictive variable is obtained to account initial data progress feature extraction.
In the specific embodiment of the present embodiment, as shown in Fig. 2, S200 is mainly comprised the steps that
S210, account initial credit ratings data, credit line behaviour in service number are obtained according to the account initial data
According to, recent refund behavioral data and the variable data of this four dimensions data of overdue behavioral data in the recent period.
Account initial credit ratings data, credit line behaviour in service number are derived by the account initial data of acquisition
According to, recent refund behavioral data and the variable data of this four dimensions data of overdue behavioral data in the recent period.It should be noted that
In the specific implementation of the present embodiment, the variable data quantity of this four dimensions data is 86, specific name such as following table institute
Show:
S220, the predictive variable is obtained to variable data progress data prediction.
Data prediction is carried out according to the variable data of above-mentioned four dimensions data, to obtain predictive variable.
In the specific embodiment of the present embodiment, as shown in figure 3, S220 is mainly comprised the steps that
S221, missing values processing and outlier processing are carried out to the variable data.
Wherein, when carrying out outlier processing to variable data, data spy is carried out to the variable data in four dimensions first
Rope, that is, variable data being described property data distribution is counted, by checking the statistical indicator of variable data, such as maximum
Value, minimum value, mean value, median etc. understand the distribution of variable data, according to the descriptive data of variable data point
Cloth statistics to carry out outlier processing to variable data；When carrying out missing values processing to variable data, mainly by miss rate
Excessively high variable is deleted.
S222, the emphasis index value for passing through the variable data of missing values processing and outlier processing is calculated.
We corresponding to each account will there is account history credit situation to establish user tag, and the account having a good credit is determined
Adopted preferably user, is defined as bad user for the bad account of credit, meanwhile, it can be according to the overdue situation of user come brief judgement
Whether variable data is obvious to the differentiation effect of fine or not user, and herein, emphasis index value is for measuring in sample data
An index of the variable data to fine or not user's separating capacity, that is to say, that the emphasis index value of variable data is bigger, prediction
Power is also bigger, we can select the IV value of variable data as emphasis index value, and the full name of IV value is information
Value, Chinese are exactly information content or the value of information, and main is exactly to screen to variable data.
In the specific implementation of the present embodiment, the IV value of the variable data in four dimensions is for example shown in following table:
It should be noted that the emphasis index value in the present invention is not limited only to IV value, there are also believe for similar emphasis index value
Cease yield value, Gini coefficient, likelihood ratio etc..
S223, the emphasis index value of the variable data is compared with preset threshold, according to comparison result to described
Variable data carries out preliminary screening.
By step S222, the emphasis index value of our available variable datas, since sample data is excessive, different accounts
Number showed with different refund, and each variable data possesses different emphasis index values, by by the weight of variable data
Point index value is compared with preset threshold, and to screen the higher variable data of predictive power, preset threshold can be according to user's need
It asks and is set, such as preset threshold can be 0.02, when carrying out preliminary screening, first by the emphasis index value of variable data
It is compared with preset threshold, when the emphasis index value of variable data is less than preset threshold, illustrates the prediction of the variable data
Power is too small, is rejected, and when the emphasis index value of variable data is greater than or equal to preset threshold, illustrate the variable number
According to predictive power it is up to standard, in such manner, it is possible to calculation amount be reduced, to improve working efficiency.
S224, the variable data after preliminary screening is grouped processing.
Being grouped processing to variable data is some categories combinations in variable data to be reduced to its radix and by variable
Data are grouped arrangement according to certain rule, that is, reach packet transaction, for example, according to variable data emphasis index value from
It is small to it is big, from big to small, U-shaped distribution or other rules are arranged.
S225, the variable data after packet transaction is subjected to the conversion of WOE evidence weight.
It is understood that the WOE value of variable data is lower herein, then represent in the grouping where the variable data
User be bad user risk it is higher.
S226, correlation analysis will be carried out by the variable data of WOE evidence weight conversion, based on the analysis results to process
The variable data of WOE evidence weight conversion carries out postsearch screening, obtains predictive variable.
Herein, correlation analysis will be carried out by the variable data of WOE evidence weight conversion first, if two variables
Correlation between data is stronger, then rejects the lesser variable data of emphasis index value, that is to say, that not by predictive power
High variable data has carried out secondary screening, to obtain final predictive variable, reduces calculation amount, improves work effect
Rate.
S300, it is modeled according to predictive variable using machine learning algorithm, obtains the pre- of response variable and predictive variable
Survey model.
When being established to prediction model, machine learning algorithm can be used and modeled, response variable can be with herein
It is set according to user, such as whether response variable can be set as to account overdue, it should be noted that we this
In the machine learning algorithm that uses may include that logistic regression algorithm, random forests algorithm or gradient are promoted in decision Tree algorithms
It is one or more.
S400, the account for having performance of refunding is predicted according to the prediction model, obtains disobeying in account time span of forecast
About probability.
In S300 step, when specifically being predicted, we will input the predictive variable for having the account showed of refunding
It is predicted in real time in prediction model, to obtain the Default Probability of account, time span of forecast here be can be understood as not
Come in a period of time.
In the specific implementation of the present embodiment, as shown in figure 4, S400 is mainly comprised the steps that
S410, prediction index is filtered out from the predictive variable.
Herein, in order to further reduce operand and promote the speed of prediction, so we can will predict
Prediction index is filtered out in variable, illustratively, the title of prediction index can be as shown in following table:
S420, it the prediction index is brought into obtains the Default Probability in account time span of forecast in the prediction model.
By obtaining account after the prediction index screened is brought into established prediction model
Default Probability in time span of forecast.
S500, according to the Default Probability in the account time span of forecast, calculate the behavior scoring of account.
Illustratively, we can obtain account in S400 step by the behavior scoring of account using 100 points of full marks as standard
After the Default Probability at family, behavior scoring, conversion formula can be converted for Default Probability by conversion formula specifically:
P=100 (1-C)
Wherein, P is behavior scoring, and C is Default Probability.
By above-mentioned conversion formula, we obtain the behavior scoring of account, in specific implementation of the invention, pass through machine
Learning algorithm is modeled, and quickly can carry out weight to prediction model compared to traditional expert's rule or scorecard model
New training carries out quick iteration update to the variable data in prediction model, quickly carries out to the behavior scoring of account pre-
It surveys and exports, increase working efficiency, the present invention is made using account credit rating data and the practical refund situation data of account
For sample data, data acquisition speed is accelerated, every transaction all the time of each account can be predicted in real time,
The applicable and generalization ability of prediction model is enhanced, includes the actual lending platforms of account in account initial data on the other hand
Refund behavioral data predicts accurate qualitative height, be conducive to company borrow in policy management work, effectively reduce financial institution
Credit risk, and the present invention will establish model and initially grade it is included, before capable of obtaining account refund behavior
The behavior scoring for evaluating account, improves prediction effect.
In the specific implementation of the present embodiment, the method also includes having to assess prediction model by appraisal procedure
The step of, wherein the appraisal procedure include: one of Lorentz curve, ROC curve, AUC statistic and KS curve or
It is a variety of.
After the building of above-mentioned prediction model, necessity will assess its accuracy, and common appraisal procedure includes Lip river
Human relations hereby curve, ROC curve (Receiver operating curve) and KS curve (Ke Ermo can love-Si meter love examine) etc.,
It can be analyzed simultaneously in conjunction with confusion matrix, promotion figure, it should be noted that the face for ROC curve, below ROC curve
Product is known as AUC statistic, and AUC value is bigger, illustrates that the resolving effect of model is better, and in KS curve, KS value is bigger, illustrates pre-
The prediction effect for surveying imperial evil spirit is better, if there is multiple prediction models, can pass through the assessment point that multiple prediction models are comprehensively compared
Value is come the prediction model that selects point value of evaluation optimal.
It should be noted that we are according to predictive variable when modeling, usually by account initial data according to certain
Ratio cut partition is training set and test set, and herein, training set is mainly used for being modeled, and test set is mainly used for building
Good prediction model is assessed, and illustratively, account initial data can be divided into training according to 7 to 3 ratio by us
Collection and test set.
Fig. 5 show another embodiment of the invention proposition a kind of Behavior-based control model loan in monitor points-scoring system,
As shown in figure 5, the system comprises:
Module is obtained, for obtaining the account initial data in the observation period；
Characteristic extracting module carries out feature extraction to the account initial data and obtains predictive variable；
Modeling module is modeled according to predictive variable, obtains the prediction model of response variable and predictive variable；
Prediction module obtains in account time span of forecast according to the prediction model to there is the account refunded and showed to predict
Default Probability；
Grading module calculates the behavior scoring of account according to the Default Probability in the account future time section.
In the specific implementation of the present embodiment, the system also includes: evaluation module, for being carried out to the prediction module
Assessment.
Yet another embodiment of the present invention provides a kind of computer equipment, including memory, processor and is stored in
On reservoir and the computer program that can run on a processor, processor realize the loan of above-mentioned Behavior-based control model when executing program
Middle monitoring methods of marking.If Fig. 6 shows, suitable for being used to realize the computer system of server provided in this embodiment, including center
Processing unit (CPU) can be loaded at random according to the program being stored in read-only memory (ROM) or from storage section
It accesses the program in memory (RAM) and executes various movements appropriate and processing.In RAM, it is also stored with computer system
Various programs and data needed for operation.CPU, ROM and RAM are connected by bus by this.Input/input (I/O) interface
It is connected to bus.
I/O interface is connected to lower component: the importation including keyboard, mouse etc.；Including such as liquid crystal display
And the output par, c of loudspeaker etc. (LCD) etc.；Storage section including hard disk etc.；And including such as LAN card, modulation /demodulation
The communications portion of the network interface card of device etc..Communications portion executes communication process via the network of such as internet.Driver
It is connected to I/O interface as needed.Detachable media, such as disk, CD, magneto-optic disk, semiconductor memory etc., according to need
It installs on a drive, in order to be mounted into storage section as needed from the computer program read thereon.
Particularly, it mentions according to the present embodiment, the process of flow chart description above may be implemented as computer software programs.Example
Such as, the present embodiment includes a kind of computer program product comprising the computer being tangibly embodied on computer-readable medium
Program, above-mentioned computer program include the program code for method shown in execution flow chart.In such embodiments, should
Computer program can be downloaded and installed from network by communications portion, and/or be mounted from detachable media.
Flow chart and schematic diagram in attached drawing, illustrate the system of the present embodiment, method and computer program product can
The architecture, function and operation being able to achieve.In this regard, each box in flow chart or schematic diagram can represent a mould
A part of block, program segment or code, a part of above-mentioned module, section or code include one or more for realizing rule
The executable instruction of fixed logic function.It should also be noted that in some implementations as replacements, function marked in the box
It can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated can actually be basic
It is performed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.It is also noted that
Each box and signal and/or the combination of the box in flow chart in schematic diagram and/or flow chart, can use and execute rule
The dedicated hardware based systems of fixed functions or operations is realized, or can use the group of specialized hardware and computer instruction
It closes to realize.
Unit involved by description in this present embodiment can be realized by way of software, can also pass through hardware
Mode is realized.Described unit also can be set in the processor, for example, can be described as: a kind of processor includes obtaining
Modulus block, computing module, detection module etc..Wherein, the title of these units is not constituted under certain conditions to the unit sheet
The restriction of body.For example, computing module is also described as " prediction module ".
As on the other hand, present invention also provides a kind of computer readable storage medium, the computer-readable storage mediums
Matter can be computer readable storage medium included in device described in above-described embodiment；It is also possible to individualism, not
The computer readable storage medium being fitted into terminal.The computer-readable recording medium storage have one or more than one
Program, described program are used to execute the loan for being described in Behavior-based control model of the invention by one or more than one processor
Middle monitoring methods of marking.
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair
The restriction of embodiments of the present invention may be used also on the basis of the above description for those of ordinary skill in the art
To make other variations or changes in different ways, all embodiments can not be exhaustive here, it is all to belong to this hair
The obvious changes or variations that bright technical solution is extended out are still in the scope of protection of the present invention.
Claims (12)
1. monitoring methods of marking in a kind of loan of Behavior-based control model, which comprises the following steps:
Obtain the account initial data in the observation period；
Feature extraction is carried out to the account initial data and obtains predictive variable；
It is modeled according to the predictive variable using machine learning algorithm, obtains the prediction mould of response variable and predictive variable
Type；
According to the prediction model to there is the account refunded and showed to predict, the Default Probability in account time span of forecast is obtained；
According to the Default Probability in the account time span of forecast, the behavior scoring of account is calculated.
2. the method according to claim 1, wherein the account initial data includes account credit rating data
And the practical refund situation data of account.
3. the method according to claim 1, wherein described obtain account initial data progress feature extraction
Include: to the step of predictive variable
According to the account initial data obtain account initial credit ratings data, credit line behaviour in service data, in the recent period also
Money behavioral data and the in the recent period variable data of this four dimensions data of overdue behavioral data；
Data prediction is carried out to the variable data and obtains the predictive variable.
4. according to the method described in claim 3, it is characterized in that, described obtain variable data progress data prediction
The step of predictive variable includes:
Missing values processing and outlier processing are carried out to the variable data；
Calculate the emphasis index value of the variable data by missing values processing and outlier processing；
The emphasis index value of the variable data is compared with preset threshold, according to comparison result to the variable data into
Row preliminary screening；
Variable data after preliminary screening is grouped processing；
Variable data after packet transaction is subjected to the conversion of WOE evidence weight；
Correlation analysis will be carried out by the variable data of WOE evidence weight conversion, based on the analysis results to by WOE weight evidence
The variable data converted again carries out postsearch screening, obtains predictive variable.
5. the method according to claim 1, wherein the title of the variable data is as shown in the table:
6. according to the method described in claim 5, it is characterized in that, it is described according to the prediction model to have refund show account
The step of family is predicted in real time, obtains the Default Probability in account time span of forecast include:
Prediction index is filtered out from the predictive variable；
The prediction index is brought into and obtains the Default Probability in account time span of forecast in the prediction model；
Wherein, the prediction index includes:
Total overdue issue,
The maximum overdue number of days of history,
The last time refunds away from modern duration
Amount can be used
History it is the last it is overdue to term day duration _ maximum,
Total loaning bill capital account for accrediting amount ratio,
Acute conjunctivitis point 1.1,
Residue should go back capital/total loaning bill capital and
Current rainbow grading.
7. the method according to claim 1, wherein further including having to comment prediction model by appraisal procedure
The step of estimating, wherein the appraisal procedure includes: one of Lorentz curve, ROC curve and KS curve or a variety of.
8. the method according to claim 1, wherein the machine learning algorithm include logistic regression algorithm, with
Machine forest algorithm or gradient promote one of decision Tree algorithms or a variety of.
9. monitoring points-scoring system in a kind of loan of Behavior-based control model characterized by comprising
Module is obtained, for obtaining the account initial data in the observation period；
Characteristic extracting module carries out feature extraction to the account initial data and obtains predictive variable；
Modeling module is modeled according to the predictive variable, obtains the prediction model of response variable and predictive variable；
Prediction module obtains disobeying in account time span of forecast according to the prediction model to there is the account refunded and showed to predict
About probability；
Grading module calculates the behavior scoring of account according to the Default Probability in the account future time section.
10. system according to claim 9, which is characterized in that further include:
Evaluation module, for assessing the prediction module.
11. a kind of computer equipment including memory, processor and stores the meter that can be run on a memory and on a processor
Calculation machine program, which is characterized in that the processor realizes such as side of any of claims 1-8 when executing described program
Method.
12. a kind of computer readable storage medium, which is characterized in that instruction is stored in the computer readable storage medium,
When the computer readable storage medium is run on computers, so that any in computer perform claim requirement 1-8
Method described in.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810966371.3A CN109191282A (en) | 2018-08-23 | 2018-08-23 | Methods of marking and system are monitored in a kind of loan of Behavior-based control model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810966371.3A CN109191282A (en) | 2018-08-23 | 2018-08-23 | Methods of marking and system are monitored in a kind of loan of Behavior-based control model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109191282A true CN109191282A (en) | 2019-01-11 |
Family
ID=64919700
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810966371.3A Pending CN109191282A (en) | 2018-08-23 | 2018-08-23 | Methods of marking and system are monitored in a kind of loan of Behavior-based control model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109191282A (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110147938A (en) * | 2019-04-23 | 2019-08-20 | 北京淇瑀信息科技有限公司 | A kind of training sample generation method, device, system and recording medium |
| CN110276369A (en) * | 2019-04-24 | 2019-09-24 | 武汉众邦银行股份有限公司 | Feature selection approach, device, equipment and storage medium based on machine learning |
| CN110276552A (en) * | 2019-06-21 | 2019-09-24 | 深圳前海微众银行股份有限公司 | Risk analysis method, device, equipment and readable storage medium storing program for executing before borrowing |
| CN110348997A (en) * | 2019-06-28 | 2019-10-18 | 北京明略软件系统有限公司 | Divide the method and device of permission |
| CN110415111A (en) * | 2019-08-01 | 2019-11-05 | 信雅达系统工程股份有限公司 | Merge the method for logistic regression credit examination & approval with expert features based on user data |
| CN110751338A (en) * | 2019-10-23 | 2020-02-04 | 贵州电网有限责任公司 | Construction and early warning method for heavy overload characteristic model of distribution transformer area |
| CN110910002A (en) * | 2019-11-15 | 2020-03-24 | 安徽海汇金融投资集团有限公司 | Account receivable default risk identification method and system |
| CN111178631A (en) * | 2019-12-30 | 2020-05-19 | 广州地理研究所 | Method and system for predicting water lettuce invasion distribution area |
| CN111275541A (en) * | 2020-01-14 | 2020-06-12 | 中信百信银行股份有限公司 | Borrower quality evaluation method and system based on multi-dimensional information, electronic device and computer readable storage medium |
| CN111324862A (en) * | 2020-02-10 | 2020-06-23 | 深圳华策辉弘科技有限公司 | Method and system for monitoring behavior in loan |
| CN111582466A (en) * | 2020-05-09 | 2020-08-25 | 深圳市卡数科技有限公司 | Scoring card configuration method, device, equipment and storage medium for simulation neural network |
| CN111738331A (en) * | 2020-06-19 | 2020-10-02 | 北京同邦卓益科技有限公司 | User classification method and device, computer-readable storage medium and electronic device |
| CN111861729A (en) * | 2020-07-31 | 2020-10-30 | 重庆富民银行股份有限公司 | Behavior scoring system and method based on lstm |
| CN112102074A (en) * | 2020-10-14 | 2020-12-18 | 深圳前海弘犀智能科技有限公司 | Grading card modeling method |
| WO2020253381A1 (en) * | 2019-06-17 | 2020-12-24 | 深圳壹账通智能科技有限公司 | Data monitoring method and apparatus, computer device and storage medium |
| CN112950354A (en) * | 2021-02-26 | 2021-06-11 | 中国光大银行股份有限公司 | Credit scoring method and device for account, storage medium and electronic device |
| CN113011624A (en) * | 2019-12-18 | 2021-06-22 | 中移(上海)信息通信科技有限公司 | User default prediction method, device, equipment and medium |
| CN111861699B (en) * | 2020-07-02 | 2021-06-22 | 北京睿知图远科技有限公司 | Anti-fraud index generation method based on operator data |
| CN114445145A (en) * | 2022-01-30 | 2022-05-06 | 中国农业银行股份有限公司 | Account prediction method and device, electronic equipment and storage medium |
-
2018
- 2018-08-23 CN CN201810966371.3A patent/CN109191282A/en active Pending
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110147938A (en) * | 2019-04-23 | 2019-08-20 | 北京淇瑀信息科技有限公司 | A kind of training sample generation method, device, system and recording medium |
| CN110276369A (en) * | 2019-04-24 | 2019-09-24 | 武汉众邦银行股份有限公司 | Feature selection approach, device, equipment and storage medium based on machine learning |
| WO2020253381A1 (en) * | 2019-06-17 | 2020-12-24 | 深圳壹账通智能科技有限公司 | Data monitoring method and apparatus, computer device and storage medium |
| CN110276552A (en) * | 2019-06-21 | 2019-09-24 | 深圳前海微众银行股份有限公司 | Risk analysis method, device, equipment and readable storage medium storing program for executing before borrowing |
| CN110348997A (en) * | 2019-06-28 | 2019-10-18 | 北京明略软件系统有限公司 | Divide the method and device of permission |
| CN110415111A (en) * | 2019-08-01 | 2019-11-05 | 信雅达系统工程股份有限公司 | Merge the method for logistic regression credit examination & approval with expert features based on user data |
| CN110751338A (en) * | 2019-10-23 | 2020-02-04 | 贵州电网有限责任公司 | Construction and early warning method for heavy overload characteristic model of distribution transformer area |
| CN110910002A (en) * | 2019-11-15 | 2020-03-24 | 安徽海汇金融投资集团有限公司 | Account receivable default risk identification method and system |
| CN113011624A (en) * | 2019-12-18 | 2021-06-22 | 中移(上海)信息通信科技有限公司 | User default prediction method, device, equipment and medium |
| CN111178631A (en) * | 2019-12-30 | 2020-05-19 | 广州地理研究所 | Method and system for predicting water lettuce invasion distribution area |
| CN111178631B (en) * | 2019-12-30 | 2024-03-29 | 广州地理研究所 | Water lettuce intrusion distribution area prediction method and system |
| CN111275541A (en) * | 2020-01-14 | 2020-06-12 | 中信百信银行股份有限公司 | Borrower quality evaluation method and system based on multi-dimensional information, electronic device and computer readable storage medium |
| CN111324862A (en) * | 2020-02-10 | 2020-06-23 | 深圳华策辉弘科技有限公司 | Method and system for monitoring behavior in loan |
| CN111582466A (en) * | 2020-05-09 | 2020-08-25 | 深圳市卡数科技有限公司 | Scoring card configuration method, device, equipment and storage medium for simulation neural network |
| CN111582466B (en) * | 2020-05-09 | 2023-09-01 | 深圳市卡数科技有限公司 | Score card configuration method, device and equipment for simulating neural network and storage medium |
| CN111738331A (en) * | 2020-06-19 | 2020-10-02 | 北京同邦卓益科技有限公司 | User classification method and device, computer-readable storage medium and electronic device |
| CN111861699B (en) * | 2020-07-02 | 2021-06-22 | 北京睿知图远科技有限公司 | Anti-fraud index generation method based on operator data |
| CN111861729A (en) * | 2020-07-31 | 2020-10-30 | 重庆富民银行股份有限公司 | Behavior scoring system and method based on lstm |
| CN112102074A (en) * | 2020-10-14 | 2020-12-18 | 深圳前海弘犀智能科技有限公司 | Grading card modeling method |
| CN112102074B (en) * | 2020-10-14 | 2024-01-30 | 深圳前海弘犀智能科技有限公司 | Score card modeling method |
| CN112950354A (en) * | 2021-02-26 | 2021-06-11 | 中国光大银行股份有限公司 | Credit scoring method and device for account, storage medium and electronic device |
| CN114445145A (en) * | 2022-01-30 | 2022-05-06 | 中国农业银行股份有限公司 | Account prediction method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109191282A (en) | Methods of marking and system are monitored in a kind of loan of Behavior-based control model | |
| US20230009149A1 (en) | System, method and computer program for underwriting and processing of loans using machine learning | |
| KR102009309B1 (en) | Management automation system for financial products and management automation method using the same | |
| CN107633265A (en) | For optimizing the data processing method and device of credit evaluation model | |
| US8355974B2 (en) | Account level liquidity charge determination | |
| CN109993652A (en) | A kind of debt-credit assessing credit risks method and device | |
| CN109961368A (en) | Data processing method and device based on machine learning | |
| CN104321794A (en) | A system and method using multi-dimensional rating to determine an entity's future commercial viability | |
| CN107633030A (en) | Credit estimation method and device based on data model | |
| CN110930038A (en) | Loan demand identification method, loan demand identification device, loan demand identification terminal and loan demand identification storage medium | |
| CN114048436A (en) | Construction method and construction device for forecasting enterprise financial data model | |
| CN107133862A (en) | Dynamic produces the method and system of the detailed transaction payment experience of enhancing credit evaluation | |
| US20110125623A1 (en) | Account level cost of funds determination | |
| Iranmanesh et al. | Customer churn prediction using artificial neural network: An analytical CRM application | |
| CN109102396A (en) | A kind of user credit ranking method, computer equipment and readable medium | |
| CN116911994A (en) | External trade risk early warning system | |
| CN117132383A (en) | Credit data processing method, device, equipment and readable storage medium | |
| CN116228403A (en) | Personal bad asset valuation method and system based on machine learning algorithm | |
| KR102499182B1 (en) | Loan regular auditing system using artificia intellicence | |
| KR102336462B1 (en) | Apparatus and method of credit rating | |
| KR102499181B1 (en) | Loan regular auditing system using artificia intellicence | |
| CN115564551A (en) | Enterprise credit rating method for financial big data | |
| CN114612239A (en) | Stock public opinion monitoring and wind control system based on algorithm, big data and artificial intelligence | |
| US8595114B2 (en) | Account level interchange effectiveness determination | |
| CN114298172A (en) | Early warning method and system based on economic value and risk prediction |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190111 |
|
| RJ01 | Rejection of invention patent application after publication |