CN108256691A - Refund Probabilistic Prediction Model construction method and device - Google Patents
Refund Probabilistic Prediction Model construction method and device Download PDFInfo
- Publication number
- CN108256691A CN108256691A CN201810126611.9A CN201810126611A CN108256691A CN 108256691 A CN108256691 A CN 108256691A CN 201810126611 A CN201810126611 A CN 201810126611A CN 108256691 A CN108256691 A CN 108256691A
- Authority
- CN
- China
- Prior art keywords
- model
- algorithm
- refund
- collection
- customer data
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
Abstract
An embodiment of the present invention provides a kind of refund Probabilistic Prediction Model construction method and devices, belong to credit prediction field, this method obtains training set of multiple promise breaking historical customer data collection as model first, feature extraction is concentrated through from the training obtain loan customer data characteristics collection again, based on the loan customer data characteristics collection and preset algorithm structure refund Probabilistic Prediction Model, the refund probability of client can be predicted by the refund Probabilistic Prediction Model of structure, so as to effectively identify the relatively low client of refund possibility, to find high risk client as early as possible, corresponding collection means collection to be taken to refund, realize the collection arrangement more rationalized, significantly improve collection efficiency.
Description
Technical field
The present invention relates to credit appraisal field, in particular to a kind of refund Probabilistic Prediction Model construction method and dress
It puts.
Background technology
China individual retail credit industry flourishes, credit card, housing loan, loan for purchasing car, personal loans for supporting students,
The fields such as durable consumer goods loan, the fields such as synthetic gold loan are flourished on line, and promise breaking client total amount expansion is matched with collection manpower
Contradiction than relative deficiency becomes increasingly conspicuous, and requirements at the higher level are proposed to the collection resource management of lending agency.
The collection business of domestic loan mechanism is mainly managed by people at present, way to manage and borrower's overdue time
It is related.In borrower's overdue early stage (within overdue 30 days), collection dynamics is smaller, is with short massage notice and simple telephone communication
Main, as the overdue time increases, collection dynamics is gradually strengthened, and when account becomes bad account, (overdue 90 days or more) can just adopt
Take the collection means of high intensity, such as collection of visiting, judicial collection.
So this collection management method based on people can lead to the problem of two：First is for high risk borrower
Early stage collection dynamics is inadequate, causes to miss the best opportunity；Second is that collection resource is disperseed, and beard eyebrow is tackled all problems at once, effect
It is undesirable.
Invention content
In view of this, the embodiment of the present invention is designed to provide a kind of refund Probabilistic Prediction Model construction method and dress
It puts, to improve the above problem.
In a first aspect, an embodiment of the present invention provides a kind of refund Probabilistic Prediction Model construction method, the method includes：
Obtain training set of multiple promise breaking historical customer data collection as model；Feature extraction is concentrated through from the training to be provided a loan
Customer data feature set；Based on the loan customer data characteristics collection and preset algorithm structure refund Probabilistic Prediction Model.
Further, refund Probabilistic Prediction Model is built based on the loan customer data characteristics collection and preset algorithm,
Including：Obtain at least two model algorithms in alternative model library, by described in loan customer data characteristics collection input at least
In two kinds of model algorithms, the result obtained is calculated according to each model algorithm and chooses a model for building the refund probabilistic forecasting
Model.
Further, at least two model algorithms in alternative model library are obtained, by the loan customer data characteristics collection
It inputs at least two model algorithm, the result obtained is calculated according to each model algorithm and chooses a model algorithm for building
The refund Probabilistic Prediction Model, including：At least two model algorithms in alternative model library are obtained, at least two mould
Type algorithm carries out model performance verification using K folding crosscheck methods；Based on model testing standard to will be examined by model performance
The model tested carries out standard test, obtains evaluation index statistics magnitude；The evaluation returned is examined to refer to according to each model criteria
Mark statistics magnitude size selection finally models the types of models used；By the mould of the loan customer data characteristics collection input selection
It is general for building the refund to calculate result one model algorithm of selection obtained according to the model algorithm for the corresponding algorithm of type type
Rate prediction model.
Further, the model algorithm in the alternative model library includes：Logistic regression algorithm, decision Tree algorithms, support
In vector machine algorithm, nearest neighbor algorithm, NB Algorithm, random forests algorithm and backpropagation neural network algorithm
At least two.
Further, the method further includes：The refund Probabilistic Prediction Model based on structure goes back loan user
Money probability is predicted.
Second aspect, an embodiment of the present invention provides a kind of refund Probabilistic Prediction Model construction device, described device includes：
Data acquisition module, for obtaining training set of multiple promise breaking historical customer data collection as model；Characteristic acquisition module,
Loan customer data characteristics collection is obtained for being concentrated through feature extraction from the training；Model creation module, for being based on
State loan customer data characteristics collection and preset algorithm structure refund Probabilistic Prediction Model.
Further, the model creation module, specifically for obtaining at least two model algorithms in alternative model library,
The loan customer data characteristics collection is inputted at least two model algorithm, the knot obtained is calculated according to each model algorithm
Fruit chooses a model for building the refund Probabilistic Prediction Model.
Further, the model creation module, including：Verification unit, for obtaining at least two in alternative model library
Kind model algorithm carries out model performance verification at least two model algorithm using K folding crosscheck methods；Indicator-specific statistics
Unit, for, to standard test will be carried out by the model that model performance is examined, obtaining evaluation index based on model testing standard
Count magnitude；Model selection unit, for examining the evaluation index returned statistics magnitude size choosing according to each model criteria
Take the types of models for finally modeling and using；Model construction unit, for by the loan customer data characteristics collection input selection
The corresponding algorithm of types of models calculates the result obtained according to the model algorithm and chooses a model algorithm for building the refund
Probabilistic Prediction Model.
Further, the model algorithm in the alternative model library includes：Logistic regression algorithm, decision Tree algorithms, support
In vector machine algorithm, nearest neighbor algorithm, NB Algorithm, random forests algorithm and backpropagation neural network algorithm
At least two.
Further, described device further includes：Probabilistic forecasting module, for the refund probabilistic forecasting mould based on structure
Type predicts the refund probability for the user that provides a loan.
The advantageous effect of the embodiment of the present invention：
An embodiment of the present invention provides a kind of refund Probabilistic Prediction Model construction method and devices, obtain multiple promise breakings first
Training set of the historical customer data collection as model, then be concentrated through feature extraction from the training and obtain loan customer data spy
Collection based on the loan customer data characteristics collection and preset algorithm structure refund Probabilistic Prediction Model, passes through going back for structure
Money Probabilistic Prediction Model can predict the refund probability of client, so as to effectively identify the relatively low client of refund possibility,
To find high risk client as early as possible, corresponding collection means collection to be taken to refund, the collection arrangement more rationalized is realized, significantly
Improve collection efficiency.
Other features and advantages of the present invention will illustrate, also, partly become from specification in subsequent specification
It is clear that by implementing understanding of the embodiment of the present invention.The purpose of the present invention and other advantages can be by saying what is write
Specifically noted structure is realized and is obtained in bright book, claims and attached drawing.
Description of the drawings
It in order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be understood that the following drawings illustrates only certain embodiments of the present invention, therefore is not construed as pair
The restriction of range, for those of ordinary skill in the art, without creative efforts, can also be according to this
A little attached drawings obtain other relevant attached drawings.
Fig. 1 shows a kind of structure diagram that can be applied to the electronic equipment in the embodiment of the present application；
Fig. 2 is a kind of flow chart of refund Probabilistic Prediction Model construction method provided in an embodiment of the present invention；
Fig. 3 is a kind of structure diagram of refund Probabilistic Prediction Model construction device provided in an embodiment of the present invention.
Specific embodiment
Below in conjunction with attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete
Ground describes, it is clear that described embodiment is only part of the embodiment of the present invention, instead of all the embodiments.Usually exist
The component of the embodiment of the present invention described and illustrated in attached drawing can be configured to arrange and design with a variety of different herein.Cause
This, the detailed description of the embodiment of the present invention to providing in the accompanying drawings is not intended to limit claimed invention below
Range, but it is merely representative of the selected embodiment of the present invention.Based on the embodiment of the present invention, those skilled in the art are not doing
Go out all other embodiments obtained under the premise of creative work, shall fall within the protection scope of the present invention.
It should be noted that：Similar label and letter represents similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, does not then need to that it is further defined and explained in subsequent attached drawing.Meanwhile the present invention's
In description, term " first ", " second " etc. are only used for distinguishing description, and it is not intended that instruction or hint relative importance.
Fig. 1 is please referred to, Fig. 1 shows a kind of structure diagram of electronic equipment 100 that can be applied in the embodiment of the present application.
Electronic equipment 100 can include refund Probabilistic Prediction Model construction device, memory 101, storage control 102, processor
103rd, Peripheral Interface 104, input-output unit 105, audio unit 106, display unit 107.
The memory 101, storage control 102, processor 103, Peripheral Interface 104, input-output unit 105, sound
Frequency unit 106,107 each element of display unit are directly or indirectly electrically connected between each other, to realize the transmission of data or friendship
Mutually.It is electrically connected for example, these elements can be realized between each other by one or more communication bus or signal wire.The refund
Probabilistic Prediction Model construction device can be stored in the storage including at least one in the form of software or firmware (firmware)
In device 101 or it is solidificated in the operating system (operating system, OS) of the refund Probabilistic Prediction Model construction device
Software function module.The processor 103 is used to perform the executable module stored in memory 101, such as the refund
The software function module or computer program that Probabilistic Prediction Model construction device includes.
Wherein, memory 101 may be, but not limited to, random access memory (Random Access Memory,
RAM), read-only memory (Read Only Memory, ROM), programmable read only memory (Programmable Read-Only
Memory, PROM), erasable read-only memory (Erasable Programmable Read-Only Memory, EPROM),
Electricallyerasable ROM (EEROM) (Electric Erasable Programmable Read-Only Memory, EEPROM) etc..
Wherein, for memory 101 for storing program, the processor 103 performs described program after execute instruction is received, aforementioned
The method performed by server that the stream process that any embodiment of the embodiment of the present invention discloses defines can be applied to processor 103
In or realized by processor 103.
Processor 103 can be a kind of IC chip, have the processing capacity of signal.Above-mentioned processor 103 can
To be general processor, including central processing unit (Central Processing Unit, abbreviation CPU), network processing unit
(Network Processor, abbreviation NP) etc.；Can also be digital signal processor (DSP), application-specific integrated circuit (ASIC),
Ready-made programmable gate array (FPGA) either other programmable logic device, discrete gate or transistor logic, discrete hard
Part component.It can realize or perform disclosed each method, step and the logic diagram in the embodiment of the present invention.General processor
Can be microprocessor or the processor 103 can also be any conventional processor etc..
Various input/output devices are coupled to processor 103 and memory 101 by the Peripheral Interface 104.At some
In embodiment, Peripheral Interface 104, processor 103 and storage control 102 can be realized in one single chip.Other one
In a little examples, they can be realized by independent chip respectively.
Input-output unit 105 is used to that user input data to be supplied to realize user and the server (or local terminal)
Interaction.The input-output unit 105 may be, but not limited to, mouse and keyboard etc..
Audio unit 106 provides a user audio interface, may include that one or more microphones, one or more raises
Sound device and voicefrequency circuit.
Display unit 107 provides an interactive interface (such as user's operation circle between the electronic equipment 100 and user
Face) or for display image data give user reference.In the present embodiment, the display unit 107 can be liquid crystal display
Or touch control display.Can be the capacitance type touch control screen or resistance for supporting single-point and multi-point touch operation if touch control display
Formula touch screen etc..Single-point and multi-point touch operation is supported to refer to that touch control display can sense on the touch control display one
Or at multiple positions simultaneously generate touch control operation, and by the touch control operation that this is sensed transfer to processor 103 carry out calculate and
Processing.
Various input/output devices are coupled to processor 103 and memory 101 by the Peripheral Interface 104.At some
In embodiment, Peripheral Interface 104, processor 103 and storage control 102 can be realized in one single chip.Other one
In a little examples, they can be realized by independent chip respectively.
Input-output unit 105 is used for the interaction that user input data is supplied to realize user and processing terminal.It is described defeated
Enter output unit 105 may be, but not limited to, mouse and keyboard etc..
It is appreciated that structure shown in FIG. 1 is only to illustrate, the electronic equipment 100 may also include more than shown in Fig. 1
Either less component or with the configuration different from shown in Fig. 1.Hardware, software may be used in each component shown in Fig. 1
Or combination is realized.
Fig. 2 is please referred to, Fig. 2 is a kind of flow of refund Probabilistic Prediction Model construction method provided in an embodiment of the present invention
Figure, described method includes following steps：
Step S110：Obtain training set of multiple promise breaking historical customer data collection as model.
The prediction of following refund situation is based on the passing analysis of borrower's historical information broken a contract, loaning bill personal data master
It is divided into two parts, first, application materials data set, another part is the data set of refund situation after making loans, that is, shows number after borrowing
According to collection.
Multiple promise breaking historical customer data collection can be obtained in advance, by multiple promise breaking historical customer data collection according to reality
It is training set and test set to stab cutting, for example, by multiple promise breaking historical customer data collection according to the time by as far as being closely ranked up,
Then using 1/3 nearest data as test set, in addition 2/3 data are used as training set for modeling.
For example, certain person bank in June, 2017 statistical number it has been found that certain length of maturity is the short term credit product A of 6 months
Generating promise breaking loan 90,000 during -2017 years on the 1st January in 2017 on August 31, (definition of promise breaking is borrower in the phase of honouring an agreement
Between at least there are a phase refund it is overdue more than 3 days).
Xian Jia banks prepare to establish refund Probabilistic Prediction Model according to this 90,000 promise breaking data samples, and prediction A products exist
The probability that promise breaking client's future refunds during in July, 2018-December, so as to promote post-loan management efficiency.
Every promise breaking loan data have 18 fields, wherein 9 be application provide a loan when information (application materials data
Collection), 9 are the information (showing data set after borrowing) refunded after making loans, and 9 fields of application information are gender, age, marriage shape
Condition, income, place province, reference scoring, the amount of the loan, loan interest rate, the intended use of the loan, 9 fields of refund information are borrower
The refund situation (whether normally refunding, be only and no two values) of each moon in 9 months after making loans.
Then this 9 complete pen promise breaking loan is sorted from the distant to the near according to practical, first 60,000 are used as training set, rear 30,000 works
For test set.
Step S120：Feature extraction, which is concentrated through, from the training obtains loan customer data characteristics collection.
Feature set is the independent variable for establishing refund Probabilistic Prediction Model, customer historical number of breaking a contract in loaning bill client
According to collection, the data in the training set obtained in step S110 cannot be directly used to model, because each loan customer may be right
A plurality of data are answered, and the information of loan customer is it is possible that there is missing, it is therefore necessary to therefrom be taken out again after being cleaned to training set
Feature is taken, ultimately generates the data set of a two-dimensional table pattern, i.e. loan customer data characteristics collection, which should
Meet the following conditions：(1) row of each loan customer and data set is one-to-one stringent mapping relations, i.e., each loan customer
A row information multiple row field is only able to find in the data set；(2) without loss of learning；(3) all data are number.
Wherein, Feature Extraction Method is：(1) information data：It is compiled for carrying out simple heat with the variable of nonnumeric description
Code processing, the mode of processing are tieed up to increase, and if gender is there are two being worth, man and female, by gender, this row becomes two row, and one is classified as gender
Man, one is classified as gender female, and for each row all only there are two being worth, a value is 1, can represent gender as man, and a value is 0,
It is female that gender, which can be represented, naturally it is also possible in turn；(2) for loss of learning, two methods can be taken：A. there is information
The row of missing directly abandons, the situation for being suitble to missing values considerably less；B. Missing Data Filling, can take the column data average value or
The mode of mode is filled；(3) there is multiple application information for loaning bill client, the method for taking data aggregate is specific as follows：It is right
In the loan customer for having multiple application information, each row application information increases by 5 row (feature), is being averaged for the column information respectively
Value, mode, maximum value, minimum value, standard deviation.
In addition, it because does not seek unity of standard, needs the result is that by being formed from the application time to the refund data of modeling time after borrowing
Its dimensionality reduction is mapped in a column vector, the method for dimensionality reduction is：Judge whether normally refund after borrower is overdue for the first time, be
1 is then returned, otherwise returns to 0, the column vector tally set that a value is 1 or 0 can be obtained accordingly, be added to software spss
Follow-up use is done in modeler.
For example, 9 fields of information of refunding are mapped as 1 field, mapping ruler (has been refunded for borrower's the present situation
And do not refund, it has refunded labeled as 1, has not refunded labeled as 0).
Wherein, above-mentioned multiple promise breaking historical customer data collection can also be subjected to discretization using spss modeler,
Then continuous data branch mailbox discretization, such as age sort according to practical, from the distant to the near then 1/3 nearest data
Test set is used as by software spss modeler cutting separation, in addition 2/3 data are used as training set for modeling.
Wherein, branch mailbox is one and the compact work of specific traffic issues, and there is no model answers, need to only abide by
Following principle：1. branch mailbox number is moderate, very few discrimination is insufficient, crosses that at most stability is strong and inconvenient management；2. combining target
Variable, branch mailbox can embody apparent trend feature, for example, for the branch mailbox at age, performance of the age below 24 years old connects very much
Closely, 25 to 32 is close, and 33-45 is approached, and more than 45 is close, so as to be divided into four casees after the age is sorted, makees 1,2,3,4 and is recorded in
From the background, 1 sample of the age less than or equal to 24 is represented, recorded successively, thus the feature discretization of continuous type.Target variable
(label), WOE (evidence warrant, weight of evidence) is a feature.Similarly other continuous type features are whole
Discretization, discrete feature such as men and women itself are also represented with number 1 and 2, and what all features of last data were presented is discrete
State.
Next the feature calculation WOE values to these discretizations are needed.According to WOE values, can also calculate IV (value of information,
Information value), IV values can be used to represent whether this feature has significance to prediction label.Rule of thumb, when
IV<When 0.02, this feature is to prediction target variable almost without help；As 0.02≤IV<When 0.1, this feature becomes prediction target
Measurer has certain help, very weak；As 0.1≤IV<When 0.3, this feature has larger help to prediction target variable；When 0.3≤
IV<When 0.5, this feature has very great help to prediction target variable tool；But as IV≤0.5, this feature has label excessively pre-
The tendency of survey, in addition, also needing to check whether to have selected have very strong causal feature with label, whether these features can be used
In prediction model.
Pass through judgement above, it is possible to filter out to modeling useful feature, reject the feature having little significance, reduce meter
It is counted as this.Then data set is rearranged, obtains feature set, in case modeling uses.
Step S130：Based on the loan customer data characteristics collection and preset algorithm structure refund Probabilistic Prediction Model.
Obtain at least two model algorithms in alternative model library, by described in loan customer data characteristics collection input extremely
In few two kinds of model algorithms, it is pre- for building the refund probability that result one model of selection obtained is calculated according to each model algorithm
Survey model.
Specifically, at least two model algorithms in alternative model library are obtained first, and K is used at least two model
It rolls over crosscheck method and carries out model performance verification.Wherein, the model algorithm in the alternative model library includes：Logistic regression is calculated
Method, decision Tree algorithms, algorithm of support vector machine, nearest neighbor algorithm, NB Algorithm, random forests algorithm and backward biography
Broadcast at least two in neural network algorithm.
Preferably, using the modeling function in spss modeler, logistic regression (Logistic Regression) is selected
Algorithm builds refund Probabilistic Prediction Model.
Logistic regression is a kind of data analysis technique widely used when predicting that target variable is discrete variable, while
It is one of more commonly used machine learning method, for estimating the possibility of certain things.Logistic recurrence fundamentally solves
Dependent variable of having determined if it were not for continuous variable what if the problem of.The dependent variable that logistic is returned can be two classification, also may be used
To be polytypic, but two classification is more commonly used, is also more prone to explain, so the most commonly used in practice is exactly two points
The logistic of class is returned.
Cross check is a kind of model performance method of calibration, and K is greater than 1 positive integer, and initial data is divided into K equal portions,
K-1 parts are randomly selected as training set, remaining 1 part collects as verification, and grader is trained with training set, generates model,
Verification collection is recycled to test the model that training obtains, and return to performance indicator, in order to reduce sampling error, needs traversal all
Training set and verification collection combination, finally take final appraisal results of the average values as model of all generation indexs.Pass through K
Folding verification, modelling effect are verified, are applied for it and rational basis is provided under actual services environment, and K values choose acquiescence
For K=5.
Model testing standard is then based on to standard test will be carried out by the model that model performance is examined, evaluation is obtained and refers to
Mark statistics magnitude.Wherein, common evaluation index Kolmogorov-Smirnov (K-S) statistic in credit scoring card field is utilized
Value weighs prediction result, and K-S is higher to show that model is stronger to the separating capacity of positive negative sample, computational methods are：Assuming that f (s
| P) be positive sample predicted value Cumulative Distribution Function, f (s | N) is Cumulative Distribution Function of the negative sample in predicted value, then has
Then the evaluation index returned statistics magnitude size selection is examined finally to model the mould used according to each model criteria
Type type, by the corresponding algorithm of types of models of the loan customer data characteristics collection input selection, according to the model algorithm meter
It calculates the result obtained and chooses a model algorithm for building the refund Probabilistic Prediction Model.I.e. using built in python language
Various algorithms are fitted training set, carry out poor verification according to K=5, can obtain the KS value highests of random forests algorithm, institute
To choose random forests algorithm as model algorithm, directly invoke the algorithm and modeled, that is, be used to generate above-mentioned refund
Probabilistic Prediction Model.
It, can also be by the test set of above-mentioned acquisition also according to above-mentioned training in order to examine the accuracy of the refund prediction model
After the relevant treatment of collection is handled, test set is input in the refund prediction model of generation, model exports a row field, should
Field is the refund probability of each user, which with test set label is compared, obtains final KS values, if KS values are more than
30%, i.e. the model can be used as refund prediction model.
For example, if logistic regression algorithm is selected to be used as refund Probabilistic Prediction Model, model evaluation can be carried out to it, then
The regression coefficient that Logic Regression Models are calculated above need to be used, regression coefficient is converted into scoring, scoring needs to meet
In a certain range, a special statistic odds may be used herein, odds, which is equal to, not to break a contract and promise breaking accounting.It can use
Following equation represents the value relationship of scoring：Score=Offset+Factor*ln (odds), Score+pdo=
Offset+Factor*ln(2*odds).Wherein pdo (points to double the odds), which is represented, increases odds
1 times needs increased score value.Offset is translated into beginning, compensation, counteracting, and Factor is translated into factor.By pdo=Factor/
Ln2 can be obtained：Factor=pdo/ln2, Offset=Score-Factor*ln (odds).It is corresponding inside last each feature
Score Scorei=Offset/n-Factor* (β 0/n+ β i*WOEi).The value of wherein Offset and Factor can be according to front
Be calculated, n represents the quantity of feature, and β i represent the coefficient of each feature, thus will obtain the discrete score value of each feature.
Prediction result is weighed using above-mentioned evaluation index Kolmogorov-Smirnov (K-S) statistics magnitudes.KS is got over
Height shows that model is stronger to the separating capacity of positive negative sample.
It is applied on test set by software spss modeler according to training set feature set and tally set processing method.It will
Test set is also divided into two parts of feature set and tally set, and the feature set of test set, sample is input to front spss one by one
In the collection Rating Model of modeler, prediction result is returned, each sample prediction result and the corresponding label of test set are done into ratio
It is right.In IBM spss modeler, the node of K-S indexs and figure is not made directly, but can pass through the group of node
Symphysis is into K-S indexs and relational graph.If K-S values are more than 30%, model can come into operation.K-S value ranges：<0.2,
It is valueless；0.2-0.4 is subjected to；0.4-0.5, well；0.5-0.6, very well；0.6-0.75 is very good；>=0.75, excessively
It is good.
Then can the refund Probabilistic Prediction Model based on structure, the refund probability for the user that provides a loan is carried out pre-
It surveys.
Refund model can also be packaged into API, be deployed in operation system, and user can initiate refund probability by operation system
Predictions request, API read borrower's request for data from system database automatically, then export what is refunded after the borrower breaks a contract
Probability, so as to which this method can predict the refund probability for the borrower that broken a contract, the bad debt risk of quantization promise breaking client, so as to help
Concentration, batch and the automatic business processing of collection task is better achieved in lending agency.
This method also has the following effects that：
(1) collection personnel's efficiency can be improved, saves collection cost：Promise breaking client is finely divided using refund probability,
Suitable strategic process and personnel can be matched, realize maximum resource utilization, effectively save operation cost and human cost.Together
When, collection personnel can effectively identify degenerate possibility or the highest client of refund possibility by scoring, collection target,
Collection is sorted and time distribution is more clear and definite, and collection means more science can significantly improve collection efficiency.
(2) arrangement of collection resource is more reasonable：The prediction level to loaning bill client is improved, can find high wind as early as possible
Dangerous client, so as to fulfill the collection arrangement more rationalized.In overdue early stage, potential high risk client is classified as collection weight
Point by taking high collection frequency, strengthens the mode of collection dynamics, such loaning bill client can be facilitated to refund as early as possible, especially when
When high risk borrower faces multiple lending agency collections, find that the mechanism of this kind of loaning bill client and collection dynamics maximum will earliest
Most it is hopeful to withdraw overdue debt.In overdue late stage, lending agency can be by limited collection resource allocation to refund
Wish on higher client, to the overdue client of no refund wish, can directly outsourcing or through judicial proceedings, avoid ineffective labor.
(3) collection means are more reasonable, promote loaning bill customer satisfaction：By obtaining the refund probability of client, for wind
The controllable borrower in danger, can take the mode for sending reminding short message or language phone to remind refund, so as to leave high-quality visitor alone
Family.
Fig. 3 is please referred to, Fig. 3 is a kind of knot of refund Probabilistic Prediction Model construction device 200 provided in an embodiment of the present invention
Structure block diagram, described device include：
Data acquisition module 210, for obtaining training set of multiple promise breaking historical customer data collection as model.
Characteristic acquisition module 220 obtains loan customer data spy for being concentrated through feature extraction from the training
Collection.
Model creation module 230 is refunded generally for being based on the loan customer data characteristics collection and preset algorithm structure
Rate prediction model.
The model creation module 230, specifically for obtaining at least two model algorithms in alternative model library, by described in
Loan customer data characteristics collection is inputted at least two model algorithm, and the result obtained is calculated according to each model algorithm and is chosen
One model is used to build the refund Probabilistic Prediction Model.
The model creation module 230, including：
Verification unit for obtaining at least two model algorithms in alternative model library, is calculated at least two model
Method carries out model performance verification using K folding crosscheck methods.
Indicator-specific statistics unit, for based on model testing standard to standard inspection will be carried out by model that model performance is examined
It tests, obtains evaluation index statistics magnitude.
Model selection unit, for the evaluation index returned statistics magnitude size being examined to choose according to each model criteria
Finally model the types of models used.
Model construction unit, for by the corresponding calculation of types of models of the loan customer data characteristics collection input selection
Method calculates the result obtained according to the model algorithm and chooses a model algorithm for building the refund Probabilistic Prediction Model.
Model algorithm in the alternative model library includes：Logistic regression algorithm, decision Tree algorithms, support vector machines are calculated
At least two in method, nearest neighbor algorithm, NB Algorithm, random forests algorithm and backpropagation neural network algorithm
Kind.
Described device further includes：
Probabilistic forecasting module, for the refund Probabilistic Prediction Model based on structure, to the refund probability for the user that provides a loan
It is predicted.
It is apparent to those skilled in the art that for convenience and simplicity of description, the device of foregoing description
Specific work process, can refer to preceding method in corresponding process, no longer excessively repeat herein.
In conclusion an embodiment of the present invention provides a kind of refund Probabilistic Prediction Model construction method and device, obtain first
Training set of multiple promise breaking historical customer data collection as model is taken, then is concentrated through feature extraction from the training and is provided a loan
Customer data feature set based on the loan customer data characteristics collection and preset algorithm structure refund Probabilistic Prediction Model, is led to
Cross structure refund Probabilistic Prediction Model the refund probability of client can be predicted, so as to effectively identify refund possibility compared with
Low client to find high risk client as early as possible, corresponding collection means collection to be taken to refund, realizes the collection more rationalized
It arranges, significantly improves collection efficiency.
In several embodiments provided herein, it should be understood that disclosed device and method can also pass through
Other modes are realized.The apparatus embodiments described above are merely exemplary, for example, flow chart and block diagram in attached drawing
Show the device of multiple embodiments according to the present invention, the architectural framework in the cards of method and computer program product,
Function and operation.In this regard, each box in flow chart or block diagram can represent the one of a module, program segment or code
Part, a part for the module, program segment or code include one or more and are used to implement holding for defined logic function
Row instruction.It should also be noted that at some as in the realization method replaced, the function that is marked in box can also be to be different from
The sequence marked in attached drawing occurs.For example, two continuous boxes can essentially perform substantially in parallel, they are sometimes
It can perform in the opposite order, this is depended on the functions involved.It is it is also noted that every in block diagram and/or flow chart
The combination of a box and the box in block diagram and/or flow chart can use function or the dedicated base of action as defined in performing
It realizes or can be realized with the combination of specialized hardware and computer instruction in the system of hardware.
In addition, each function module in each embodiment of the present invention can integrate to form an independent portion
Point or modules individualism, can also two or more modules be integrated to form an independent part.
If the function is realized in the form of software function module and is independent product sale or in use, can be with
It is stored in a computer read/write memory medium.Based on such understanding, technical scheme of the present invention is substantially in other words
The part contribute to the prior art or the part of the technical solution can be embodied in the form of software product, the meter
Calculation machine software product is stored in a storage medium, is used including some instructions so that a computer equipment (can be
People's computer, server or network equipment etc.) perform all or part of the steps of the method according to each embodiment of the present invention.
And aforementioned storage medium includes：USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), arbitrary access are deposited
The various media that can store program code such as reservoir (RAM, Random Access Memory), magnetic disc or CD.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, that is made any repaiies
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.It should be noted that：Similar label and letter exists
Similar terms are represented in following attached drawing, therefore, once being defined in a certain Xiang Yi attached drawing, are then not required in subsequent attached drawing
It is further defined and is explained.
The above description is merely a specific embodiment, but protection scope of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in change or replacement, should all contain
Lid is within protection scope of the present invention.Therefore, protection scope of the present invention described should be subject to the protection scope in claims.
It should be noted that herein, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any this practical relationship or sequence.Moreover, term " comprising ", "comprising" or its any other variant are intended to
Non-exclusive inclusion, so that process, method, article or equipment including a series of elements not only will including those
Element, but also including other elements that are not explicitly listed or further include as this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
Also there are other identical elements in process, method, article or equipment including the element.
Claims (10)
1. a kind of refund Probabilistic Prediction Model construction method, which is characterized in that the method includes：
Obtain training set of multiple promise breaking historical customer data collection as model；
Feature extraction, which is concentrated through, from the training obtains loan customer data characteristics collection；
Based on the loan customer data characteristics collection and preset algorithm structure refund Probabilistic Prediction Model.
2. according to the method described in claim 1, it is characterized in that, based on the loan customer data characteristics collection and pre- imputation
Method builds refund Probabilistic Prediction Model, including：
At least two model algorithms in alternative model library are obtained, it will be at least two described in loan customer data characteristics collection input
In kind model algorithm, the result obtained is calculated according to each model algorithm and chooses a model for building the refund probabilistic forecasting mould
Type.
3. according to the method described in claim 2, it is characterized in that, obtain at least two model algorithms in alternative model library,
The loan customer data characteristics collection is inputted at least two model algorithm, the knot obtained is calculated according to each model algorithm
Fruit chooses a model algorithm for building the refund Probabilistic Prediction Model, including：
At least two model algorithms in alternative model library are obtained, at least two model algorithm using K folding crosschecks
Method carries out model performance verification；
Based on model testing standard to standard test will be carried out by the model that model performance is examined, evaluation index statistic is obtained
Value；
The evaluation index returned statistics magnitude size selection is examined finally to model the model class used according to each model criteria
Type；
By the corresponding algorithm of types of models of the loan customer data characteristics collection input selection, obtained according to model algorithm calculating
The result obtained chooses a model algorithm for building the refund Probabilistic Prediction Model.
4. according to the method described in claim 3, it is characterized in that, the model algorithm in the alternative model library includes：Logic
Regression algorithm, decision Tree algorithms, algorithm of support vector machine, nearest neighbor algorithm, NB Algorithm, random forests algorithm and
At least two in backpropagation neural network algorithm.
5. according to the method described in claim 1, it is characterized in that, the method further includes：
The refund Probabilistic Prediction Model based on structure predicts the refund probability for the user that provides a loan.
6. a kind of refund Probabilistic Prediction Model construction device, which is characterized in that described device includes：
Data acquisition module, for obtaining training set of multiple promise breaking historical customer data collection as model；
Characteristic acquisition module obtains loan customer data characteristics collection for being concentrated through feature extraction from the training；
Model creation module, for being based on the loan customer data characteristics collection and preset algorithm structure refund probabilistic forecasting mould
Type.
7. device according to claim 6, which is characterized in that the model creation module, specifically for obtaining alternative mould
At least two model algorithms in type library input the loan customer data characteristics collection at least two model algorithm,
The result obtained is calculated according to each model algorithm and chooses a model for building the refund Probabilistic Prediction Model.
8. device according to claim 7, which is characterized in that the model creation module, including：
Verification unit for obtaining at least two model algorithms in alternative model library, adopts at least two model algorithm
Model performance verification is carried out with K folding crosscheck methods；
Indicator-specific statistics unit, for based on model testing standard to standard test will be carried out by model that model performance is examined,
Obtain evaluation index statistics magnitude；
Model selection unit, it is final for the evaluation index returned statistics magnitude size being examined to choose according to each model criteria
Model the types of models used；
Model construction unit, for by the corresponding algorithm of types of models of the loan customer data characteristics collection input selection, root
The result obtained is calculated according to the model algorithm and chooses a model algorithm for building the refund Probabilistic Prediction Model.
9. device according to claim 8, which is characterized in that the model algorithm in the alternative model library includes：Logic
Regression algorithm, decision Tree algorithms, algorithm of support vector machine, nearest neighbor algorithm, NB Algorithm, random forests algorithm and
At least two in backpropagation neural network algorithm.
10. device according to claim 6, which is characterized in that described device further includes：
Probabilistic forecasting module for the refund Probabilistic Prediction Model based on structure, carries out the refund probability for the user that provides a loan
Prediction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810126611.9A CN108256691A (en) | 2018-02-08 | 2018-02-08 | Refund Probabilistic Prediction Model construction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810126611.9A CN108256691A (en) | 2018-02-08 | 2018-02-08 | Refund Probabilistic Prediction Model construction method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108256691A true CN108256691A (en) | 2018-07-06 |
Family
ID=62744134
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810126611.9A Pending CN108256691A (en) | 2018-02-08 | 2018-02-08 | Refund Probabilistic Prediction Model construction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108256691A (en) |
Cited By (34)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109360084A (en) * | 2018-09-27 | 2019-02-19 | 平安科技（深圳）有限公司 | Appraisal procedure and device, storage medium, the computer equipment of reference default risk |
| CN109522317A (en) * | 2018-09-05 | 2019-03-26 | 深圳市佰仟金融服务有限公司 | A kind of anti-fraud method for early warning and system |
| CN109657837A (en) * | 2018-11-19 | 2019-04-19 | 平安科技（深圳）有限公司 | Default Probability prediction technique, device, computer equipment and storage medium |
| CN109685336A (en) * | 2018-12-10 | 2019-04-26 | 深圳市小牛普惠投资管理有限公司 | Collection task distribution method, device, computer equipment and storage medium |
| CN109710890A (en) * | 2018-12-20 | 2019-05-03 | 四川新网银行股份有限公司 | Behavior portrait model based on building identifies the method and system of false material in real time |
| CN109933834A (en) * | 2018-12-26 | 2019-06-25 | 阿里巴巴集团控股有限公司 | A kind of model creation method and device of time series data prediction |
| CN110033165A (en) * | 2019-03-06 | 2019-07-19 | 平安科技（深圳）有限公司 | The recommended method of overdue loaning bill collection mode, device, medium, electronic equipment |
| CN110046986A (en) * | 2019-03-06 | 2019-07-23 | 平安科技（深圳）有限公司 | The overdue customer grouping method and device of loaning bill based on big data |
| CN110245985A (en) * | 2019-06-11 | 2019-09-17 | 深圳前海微众银行股份有限公司 | A kind of information processing method and device |
| CN110276677A (en) * | 2019-04-24 | 2019-09-24 | 武汉众邦银行股份有限公司 | Refund prediction technique, device, equipment and storage medium based on big data platform |
| CN110288460A (en) * | 2019-04-24 | 2019-09-27 | 武汉众邦银行股份有限公司 | Collection prediction technique, device, equipment and storage medium based on propagated forward |
| CN110363658A (en) * | 2019-07-16 | 2019-10-22 | 北京明略软件系统有限公司 | Processing method and processing device, storage medium and the electronic device of credit data |
| CN110363650A (en) * | 2019-06-27 | 2019-10-22 | 上海淇毓信息科技有限公司 | A kind of storage user dynamic branch wish prediction technique, device and system |
| CN110688373A (en) * | 2019-09-17 | 2020-01-14 | 杭州绿度信息技术有限公司 | OFFSET method based on logistic regression |
| CN110738564A (en) * | 2019-10-16 | 2020-01-31 | 信雅达系统工程股份有限公司 | Post-loan risk assessment method and device and storage medium |
| WO2020024448A1 (en) * | 2018-08-01 | 2020-02-06 | 平安科技（深圳）有限公司 | Group performance grade identification method, device, storage medium, and computer apparatus |
| CN111062518A (en) * | 2019-11-22 | 2020-04-24 | 成都铂锡金融信息技术有限公司 | Method, device and storage medium for processing hastening service based on artificial intelligence |
| CN111199469A (en) * | 2019-12-12 | 2020-05-26 | 北京淇瑀信息科技有限公司 | User payment model generation method and device and electronic equipment |
| CN111222979A (en) * | 2019-12-27 | 2020-06-02 | 安徽科讯金服科技有限公司 | Loan credit evaluation system based on government affair big data |
| CN111369336A (en) * | 2020-02-21 | 2020-07-03 | 四川新网银行股份有限公司 | Method for prompting borrowing in bank |
| CN111681102A (en) * | 2020-06-05 | 2020-09-18 | 深圳市卡牛科技有限公司 | Credit prediction method, apparatus, device and storage medium |
| CN111709828A (en) * | 2020-06-12 | 2020-09-25 | 中国建设银行股份有限公司 | Resource processing method, device, equipment and system |
| CN112035582A (en) * | 2020-08-28 | 2020-12-04 | 光大科技有限公司 | Structured data classification method and device, storage medium and electronic device |
| CN112308295A (en) * | 2020-10-10 | 2021-02-02 | 北京贝壳时代网络科技有限公司 | Default probability prediction method and device |
| CN112330476A (en) * | 2020-11-27 | 2021-02-05 | 中国人寿保险股份有限公司 | Method and device for predicting group insurance business |
| CN112330047A (en) * | 2020-11-18 | 2021-02-05 | 交通银行股份有限公司 | Credit card repayment probability prediction method based on user behavior characteristics |
| CN112785418A (en) * | 2021-01-22 | 2021-05-11 | 深圳前海微众银行股份有限公司 | Credit risk modeling method, credit risk modeling device, credit risk modeling equipment and computer readable storage medium |
| CN112819613A (en) * | 2021-03-09 | 2021-05-18 | 重庆度小满优扬科技有限公司 | Loan information processing method, loan information processing apparatus, and storage medium |
| 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 |
| CN113065945A (en) * | 2021-03-17 | 2021-07-02 | 上海浦东发展银行股份有限公司 | Method and system for classifying repayment willingness of customer for collection, verification and sale |
| CN113298510A (en) * | 2018-07-10 | 2021-08-24 | 马上消费金融股份有限公司 | Deduction instruction initiating method and device |
| CN113657901A (en) * | 2021-07-23 | 2021-11-16 | 上海钧正网络科技有限公司 | Method, system, terminal and medium for managing collection of owing user |
| CN113822490A (en) * | 2021-09-29 | 2021-12-21 | 平安银行股份有限公司 | Asset clearing and accepting method and device based on artificial intelligence and electronic equipment |
| CN113822755A (en) * | 2021-09-27 | 2021-12-21 | 武汉众邦银行股份有限公司 | Method for identifying credit risk of individual user by using feature discretization technology |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8355974B2 (en) * | 2009-11-20 | 2013-01-15 | Bank Of America Corporation | Account level liquidity charge determination |
| CN105005575A (en) * | 2015-03-05 | 2015-10-28 | 张良均 | Quick developing interface method for enterprise intelligent prediction |
| CN106897918A (en) * | 2017-02-24 | 2017-06-27 | 上海易贷网金融信息服务有限公司 | A kind of hybrid machine learning credit scoring model construction method |
| CN107633455A (en) * | 2017-09-04 | 2018-01-26 | 深圳市华傲数据技术有限公司 | Credit estimation method and device based on data model |
-
2018
- 2018-02-08 CN CN201810126611.9A patent/CN108256691A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8355974B2 (en) * | 2009-11-20 | 2013-01-15 | Bank Of America Corporation | Account level liquidity charge determination |
| CN105005575A (en) * | 2015-03-05 | 2015-10-28 | 张良均 | Quick developing interface method for enterprise intelligent prediction |
| CN106897918A (en) * | 2017-02-24 | 2017-06-27 | 上海易贷网金融信息服务有限公司 | A kind of hybrid machine learning credit scoring model construction method |
| CN107633455A (en) * | 2017-09-04 | 2018-01-26 | 深圳市华傲数据技术有限公司 | Credit estimation method and device based on data model |
Cited By (42)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113298510A (en) * | 2018-07-10 | 2021-08-24 | 马上消费金融股份有限公司 | Deduction instruction initiating method and device |
| WO2020024448A1 (en) * | 2018-08-01 | 2020-02-06 | 平安科技（深圳）有限公司 | Group performance grade identification method, device, storage medium, and computer apparatus |
| CN109522317A (en) * | 2018-09-05 | 2019-03-26 | 深圳市佰仟金融服务有限公司 | A kind of anti-fraud method for early warning and system |
| CN109360084A (en) * | 2018-09-27 | 2019-02-19 | 平安科技（深圳）有限公司 | Appraisal procedure and device, storage medium, the computer equipment of reference default risk |
| CN109657837A (en) * | 2018-11-19 | 2019-04-19 | 平安科技（深圳）有限公司 | Default Probability prediction technique, device, computer equipment and storage medium |
| CN109685336A (en) * | 2018-12-10 | 2019-04-26 | 深圳市小牛普惠投资管理有限公司 | Collection task distribution method, device, computer equipment and storage medium |
| CN109710890A (en) * | 2018-12-20 | 2019-05-03 | 四川新网银行股份有限公司 | Behavior portrait model based on building identifies the method and system of false material in real time |
| CN109710890B (en) * | 2018-12-20 | 2023-06-09 | 四川新网银行股份有限公司 | Method and system for identifying false material in real time based on constructed behavior portrait model |
| CN109933834A (en) * | 2018-12-26 | 2019-06-25 | 阿里巴巴集团控股有限公司 | A kind of model creation method and device of time series data prediction |
| CN109933834B (en) * | 2018-12-26 | 2023-06-27 | 创新先进技术有限公司 | Model creation method and device for time sequence data prediction |
| CN110033165A (en) * | 2019-03-06 | 2019-07-19 | 平安科技（深圳）有限公司 | The recommended method of overdue loaning bill collection mode, device, medium, electronic equipment |
| CN110046986A (en) * | 2019-03-06 | 2019-07-23 | 平安科技（深圳）有限公司 | The overdue customer grouping method and device of loaning bill based on big data |
| CN110288460A (en) * | 2019-04-24 | 2019-09-27 | 武汉众邦银行股份有限公司 | Collection prediction technique, device, equipment and storage medium based on propagated forward |
| CN110276677A (en) * | 2019-04-24 | 2019-09-24 | 武汉众邦银行股份有限公司 | Refund prediction technique, device, equipment and storage medium based on big data platform |
| CN110245985B (en) * | 2019-06-11 | 2023-09-08 | 深圳前海微众银行股份有限公司 | Information processing method and device |
| CN110245985A (en) * | 2019-06-11 | 2019-09-17 | 深圳前海微众银行股份有限公司 | A kind of information processing method and device |
| CN110363650A (en) * | 2019-06-27 | 2019-10-22 | 上海淇毓信息科技有限公司 | A kind of storage user dynamic branch wish prediction technique, device and system |
| CN110363658A (en) * | 2019-07-16 | 2019-10-22 | 北京明略软件系统有限公司 | Processing method and processing device, storage medium and the electronic device of credit data |
| CN110688373A (en) * | 2019-09-17 | 2020-01-14 | 杭州绿度信息技术有限公司 | OFFSET method based on logistic regression |
| CN110738564A (en) * | 2019-10-16 | 2020-01-31 | 信雅达系统工程股份有限公司 | Post-loan risk assessment method and device and storage medium |
| CN111062518A (en) * | 2019-11-22 | 2020-04-24 | 成都铂锡金融信息技术有限公司 | Method, device and storage medium for processing hastening service based on artificial intelligence |
| CN111199469A (en) * | 2019-12-12 | 2020-05-26 | 北京淇瑀信息科技有限公司 | User payment model generation method and device and electronic equipment |
| 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 |
| CN111222979A (en) * | 2019-12-27 | 2020-06-02 | 安徽科讯金服科技有限公司 | Loan credit evaluation system based on government affair big data |
| CN111369336A (en) * | 2020-02-21 | 2020-07-03 | 四川新网银行股份有限公司 | Method for prompting borrowing in bank |
| CN111681102A (en) * | 2020-06-05 | 2020-09-18 | 深圳市卡牛科技有限公司 | Credit prediction method, apparatus, device and storage medium |
| CN111681102B (en) * | 2020-06-05 | 2023-09-01 | 深圳市卡牛科技有限公司 | Credit prediction method, apparatus, device and storage medium |
| CN111709828A (en) * | 2020-06-12 | 2020-09-25 | 中国建设银行股份有限公司 | Resource processing method, device, equipment and system |
| CN112035582A (en) * | 2020-08-28 | 2020-12-04 | 光大科技有限公司 | Structured data classification method and device, storage medium and electronic device |
| CN112308295A (en) * | 2020-10-10 | 2021-02-02 | 北京贝壳时代网络科技有限公司 | Default probability prediction method and device |
| CN112330047A (en) * | 2020-11-18 | 2021-02-05 | 交通银行股份有限公司 | Credit card repayment probability prediction method based on user behavior characteristics |
| CN112330476A (en) * | 2020-11-27 | 2021-02-05 | 中国人寿保险股份有限公司 | Method and device for predicting group insurance business |
| CN112785418A (en) * | 2021-01-22 | 2021-05-11 | 深圳前海微众银行股份有限公司 | Credit risk modeling method, credit risk modeling device, credit risk modeling equipment and computer readable storage medium |
| CN112785418B (en) * | 2021-01-22 | 2024-02-06 | 深圳前海微众银行股份有限公司 | Credit risk modeling method, apparatus, device and computer readable storage medium |
| CN112819613B (en) * | 2021-03-09 | 2023-07-04 | 重庆度小满优扬科技有限公司 | Loan information processing method, device and storage medium |
| CN112819613A (en) * | 2021-03-09 | 2021-05-18 | 重庆度小满优扬科技有限公司 | Loan information processing method, loan information processing apparatus, and storage medium |
| CN113065945A (en) * | 2021-03-17 | 2021-07-02 | 上海浦东发展银行股份有限公司 | Method and system for classifying repayment willingness of customer for collection, verification and sale |
| CN113657901A (en) * | 2021-07-23 | 2021-11-16 | 上海钧正网络科技有限公司 | Method, system, terminal and medium for managing collection of owing user |
| CN113657901B (en) * | 2021-07-23 | 2024-04-16 | 上海钧正网络科技有限公司 | Method, system, terminal and medium for managing fee owed users |
| CN113822755A (en) * | 2021-09-27 | 2021-12-21 | 武汉众邦银行股份有限公司 | Method for identifying credit risk of individual user by using feature discretization technology |
| CN113822755B (en) * | 2021-09-27 | 2023-09-05 | 武汉众邦银行股份有限公司 | Identification method of credit risk of individual user by feature discretization technology |
| CN113822490A (en) * | 2021-09-29 | 2021-12-21 | 平安银行股份有限公司 | Asset clearing and accepting method and device based on artificial intelligence and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108256691A (en) | Refund Probabilistic Prediction Model construction method and device | |
| CN110837931B (en) | Customer churn prediction method, device and storage medium | |
| CN108648074A (en) | Loan valuation method, apparatus based on support vector machines and equipment | |
| CN104321794B (en) | A kind of system and method that the following commercial viability of an entity is determined using multidimensional grading | |
| CN106897918A (en) | A kind of hybrid machine learning credit scoring model construction method | |
| CN106952155A (en) | A kind of collection method and device based on credit scoring | |
| CN107424070A (en) | A kind of loan user credit ranking method and system based on machine learning | |
| CN106600369A (en) | Real-time recommendation system and method of financial products of banks based on Naive Bayesian classification | |
| CN108596495A (en) | A kind of retail credit business points-scoring system and method | |
| CN110288459A (en) | Loan prediction technique, device, equipment and storage medium | |
| CN108388974A (en) | Top-tier customer Optimum Identification Method and device based on random forest and decision tree | |
| CN108389069A (en) | Top-tier customer recognition methods based on random forest and logistic regression and device | |
| CN107622326A (en) | User's classification, available resources Forecasting Methodology, device and equipment | |
| CN110288460A (en) | Collection prediction technique, device, equipment and storage medium based on propagated forward | |
| CN108154311A (en) | Top-tier customer recognition methods and device based on random forest and decision tree | |
| KR20210137604A (en) | Automatic data analysis method and system using artificial intelligence | |
| CN110276677A (en) | Refund prediction technique, device, equipment and storage medium based on big data platform | |
| CN108364191A (en) | Top-tier customer Optimum Identification Method and device based on random forest and logistic regression | |
| CN111709826A (en) | Target information determination method and device | |
| CN110930218A (en) | Method and device for identifying fraudulent customer and electronic equipment | |
| CN113034046A (en) | Data risk metering method and device, electronic equipment and storage medium | |
| CN113642923A (en) | Bad asset pack value evaluation method based on historical collection urging data | |
| CN115545886A (en) | Overdue risk identification method, overdue risk identification device, overdue risk identification equipment and storage medium | |
| CN109858947A (en) | Retail user value analysis system and method | |
| CN109191185A (en) | A kind of visitor's heap sort method and system |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180706 |