CN110276677A - Refund prediction technique, device, equipment and storage medium based on big data platform - Google Patents
Refund prediction technique, device, equipment and storage medium based on big data platform Download PDFInfo
- Publication number
- CN110276677A CN110276677A CN201910342059.1A CN201910342059A CN110276677A CN 110276677 A CN110276677 A CN 110276677A CN 201910342059 A CN201910342059 A CN 201910342059A CN 110276677 A CN110276677 A CN 110276677A
- Authority
- CN
- China
- Prior art keywords
- information
- refund
- big data
- data platform
- target
- 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
The invention discloses a kind of refund prediction technique, device, equipment and storage medium based on big data platform, which comprises requested by obtaining the loan application that user initiates in big data platform, the loan application request includes identity label information;Corresponding loan transaction information is searched in the big data platform according to the identity label information;Extract the current transaction feature information in the loan transaction information;Preset rules information is obtained, the current transaction feature information is extended according to the preset rules information, obtains target transaction characteristic information;Refund probabilistic forecasting is carried out by the default credit prediction model according to the target transaction characteristic information, obtain the target refund probability results of the user, to be extended according to preset rules information to current transaction feature information, obtain more fully predicted characteristics information, and refund probabilistic forecasting is carried out by default credit prediction model, to improve the accuracy for prediction of refunding.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of refund prediction techniques based on big data platform, dress
It sets, equipment and storage medium.
Background technique
Technology and air control are the Two Mainstays of credit operation on line, and technology is the support of bottom, and risk is the building on upper layer.
Big data at present, artificial intelligence technology is widely used in air control field, and from regulation engine to model platform, real time monitoring fraud is handed over
Easily, effective prevention and control risk of fraud.One complete air control platform needs includes being managed to loan application Life cycle, is
One extremely complex process, each process can influence whole air control quality, but carry out credit in big data platform
The data analysis that often cannot achieve more various dimensions when data processing, to cause precision of prediction not high.
Summary of the invention
It is a primary object of the present invention to propose a kind of refund prediction technique based on big data platform, device, equipment and
Storage medium, it is intended to which solve the information content when big data platform carries out data processing does not lead to the skill for predicting that processing accuracy is not high entirely
Art problem.
To achieve the above object, the present invention provides a kind of refund prediction technique based on big data platform, described based on big
The refund prediction technique of data platform the following steps are included:
Obtain the loan application request that user initiates in big data platform, wherein the loan application request includes body
Part label information;
Corresponding loan is searched in the big data platform according to the identity label information in loan application request
Transaction Information;
Extract the current transaction feature information in the loan transaction information;
Preset rules information is obtained, the current transaction feature information is extended according to the preset rules information,
Obtain target transaction characteristic information;
Refund probabilistic forecasting is carried out by the default credit prediction model according to the target transaction characteristic information, is obtained
The target refund probability results of the user.
Preferably, before the loan application request that the acquisition user initiates in big data platform, the method is also wrapped
It includes:
Obtain the connection status with the big data platform;
When the connection status is to connect normal, the loan Shen for obtaining user and initiating in big data platform is executed
The step of please requesting.
Preferably, the preset rules information includes default exhibition opening information;
The acquisition preset rules information expands the current transaction feature information according to the preset rules information
Exhibition, obtains target transaction characteristic information, comprising:
Default exhibition opening information is obtained, the current transaction feature information is expanded according to the default exhibition opening information
Exhibition, obtains target transaction characteristic information.
Preferably, default exhibition opening information is obtained, is worked as by following formula to described according to the default exhibition opening information
Preceding transaction feature information is extended, and obtains target transaction characteristic information;
The formula are as follows:
Wherein, index a, b, c, m are full
The combination of sufficient a+b+c+m≤n, and i≤j≤k≤l);
Wherein, the J indicates that target transaction characteristic information, the m indicate current dimension, and n indicates default exhibition opening information,
A, b, c and m indicate that the index information of default item, x indicate extension feature information, and i, j, k and l respectively indicate target transaction spy
Reference ceases corresponding item sequence number.
Preferably, described refund generally by the default credit prediction model according to the target transaction characteristic information
Rate prediction, before obtaining the target refund probability results of the user, the method also includes:
Obtain historical trading characteristic information and corresponding history target refund probabilistic information;
By the historical trading characteristic information composition characteristic data set, and using the characteristic data set as node;
The default historical trading characteristic information that the characteristic is concentrated is extracted, and obtains default historical trading characteristic information
The characteristic data set is divided into according to the history target refund probabilistic information by corresponding history target refund probabilistic information
One data set and the second data set;
It is obtained according to first data set and the second data set with reference to gini index;
From described with reference to the corresponding historical trading characteristic information of selection preset reference gini index and history in gini index
Target refund probabilistic information is as child node;
When the child node meets preset condition, the default credit is obtained according to the node and child node and is predicted
Model.
Preferably, described when the child node meets preset condition, it is obtained according to the node and child node described
Default credit prediction model, comprising:
When the child node meets preset condition, reference credit is generated according to the node and child node and predicts mould
Type;
Obtain sample transaction feature information, by reference credit prediction model described in the sample transaction feature information input into
Row refund probabilistic forecasting obtains prediction refund probability results;
The corresponding sample refund probability results of sample transaction feature information are obtained, by the sample refund probability results and institute
It states prediction refund probability results to be compared, obtains objective appraisal index;
Target credit prediction model is selected from the reference credit prediction model according to the objective appraisal index, it will
The target credit prediction model is as the default credit prediction model.
Preferably, the loan application request includes user's shroff account number information;
It is described that refund probabilistic forecasting is carried out by the default credit prediction model according to the target transaction characteristic information,
After obtaining the target refund probability results of the user, the method also includes:
The target refund probability results are compared with preset threshold;
When the target refund probability results are more than preset threshold, user's gathering in the loan application request is obtained
Account information is requested according to loan application described in user's shroff account number information response.
In addition, to achieve the above object, the present invention also proposes a kind of refund prediction meanss based on big data platform, described
Refund prediction meanss based on big data platform include:
Module is obtained, the loan application request initiated in big data platform for obtaining user, wherein the loan Shen
It please request to include identity label information;
Searching module, for being looked into the big data platform according to the identity label information in loan application request
Look for corresponding loan transaction information;
Extraction module, for extracting the current transaction feature information in the loan transaction information;
Expansion module, for obtaining preset rules information, according to the preset rules information to the current transaction feature
Information is extended, and obtains target transaction characteristic information;
Prediction module, for being refunded according to the target transaction characteristic information by the default credit prediction model
Probabilistic forecasting obtains the target refund probability results of the user.
In addition, to achieve the above object, the present invention also proposes a kind of pre- measurement equipment of the refund based on big data platform, described
The pre- measurement equipment of refund based on big data platform includes: memory, processor and is stored on the memory and can be described
The refund Prediction program based on big data platform run on processor, the refund Prediction program based on big data platform are matched
It is set to the step of realizing refund prediction technique based on big data platform as described above.
In addition, to achieve the above object, the present invention also proposes a kind of storage medium, it is stored with and is based on the storage medium
The refund Prediction program of big data platform is realized such as when the refund Prediction program based on big data platform is executed by processor
The step of refund prediction technique based on big data platform described above.
Refund prediction technique proposed by the present invention based on big data platform is sent out in big data platform by obtaining user
The loan application request risen, wherein the loan application request includes identity label information;According in loan application request
Identity label information corresponding loan transaction information is searched in the big data platform;It extracts in the loan transaction information
Current transaction feature information;Preset rules information is obtained, the current transaction feature is believed according to the preset rules information
Breath is extended, and obtains target transaction characteristic information;It is predicted according to the target transaction characteristic information by the default credit
Model carries out refund probabilistic forecasting, obtains the target refund probability results of the user, thus according to the preset rules information
The current transaction feature information is extended, obtains more fully predicted characteristics information, and mould is predicted by default credit
Type carries out refund probabilistic forecasting, to improve the accuracy for prediction of refunding.
Detailed description of the invention
Fig. 1 is the pre- measurement equipment of the refund based on big data platform for the hardware running environment that the embodiment of the present invention is related to
Structural schematic diagram;
Fig. 2 is that the present invention is based on the flow diagrams of the refund prediction technique first embodiment of big data platform;
Fig. 3 is that the present invention is based on the making loans for one embodiment of refund prediction technique of big data platform to predict flow diagram;
Fig. 4 is that the present invention is based on the flow diagrams of the refund prediction technique second embodiment of big data platform;
Fig. 5 is that the present invention is based on the flow diagrams of the refund prediction technique 3rd embodiment of big data platform;
Fig. 6 is that the present invention is based on the functional block diagrams of the refund prediction meanss first embodiment of big data platform.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
Referring to Fig.1, Fig. 1 is the refund based on big data platform for the hardware running environment that the embodiment of the present invention is related to
Predict device structure schematic diagram.
As shown in Figure 1, being somebody's turn to do the pre- measurement equipment of refund based on big data platform may include: processor 1001, such as center
Processor (Central Processing Unit, CPU), communication bus 1002, user interface 1003, network interface 1004 are deposited
Reservoir 1005.Wherein, communication bus 1002 is for realizing the connection communication between these components.User interface 1003 may include
Display screen (Display), input unit such as key, optional user interface 1003 can also include wireline interface, the nothing of standard
Line interface.Network interface 1004 optionally may include standard wireline interface and wireless interface (such as WI-FI interface).Memory
1005 can be high-speed random access memory (Random Access Memory, RAM) memory, be also possible to stable deposit
Reservoir (non-volatile memory), such as magnetic disk storage.Memory 1005 optionally can also be independently of aforementioned place
Manage the storage device of device 1001.
It will be understood by those skilled in the art that device structure shown in Fig. 1 is not constituted to based on big data platform
It refunds the restriction of pre- measurement equipment, may include perhaps combining certain components or different than illustrating more or fewer components
Component layout.
As shown in Figure 1, as may include operating system, network communication mould in a kind of memory 1005 of storage medium
Block, Subscriber Interface Module SIM and the refund Prediction program based on big data platform.
In the pre- measurement equipment of refund based on big data platform shown in Fig. 1, it is outer that network interface 1004 is mainly used for connection
Net carries out data communication with other network equipments;User interface 1003 is mainly used for connecting user equipment, with the user equipment
Carry out data communication;Present device calls the going back based on big data platform stored in memory 1005 by processor 1001
Money Prediction program, and execute the implementation method of the refund prediction provided in an embodiment of the present invention based on big data platform.
Based on above-mentioned hardware configuration, propose that the present invention is based on the refund prediction technique embodiments of big data platform.
It is that the present invention is based on the signals of the process of the refund prediction technique first embodiment of big data platform referring to Fig. 2, Fig. 2
Figure.
In the first embodiment, the refund prediction technique based on big data platform the following steps are included:
Step S10 obtains the loan application request that user initiates in big data platform, wherein the loan application is asked
It asks including identity label information.
It should be noted that the executing subject of the present embodiment is the pre- measurement equipment of refund based on big data platform, such as also
Money predictive server etc., can also be other equipment, the present embodiment to this with no restriction.
In the present embodiment, risk of fraud information data system, including thousands of category informations have been built based on big data technology
Content and nearly 10,000,000,000 transaction records cover client-related information, account information, card image, authorization message, clearance letter
Breath, refund information and have been acknowledged transaction swindling record etc. data, to obtain the Transaction Information of user.
In the concrete realization, apply for user Ke Tong big data platform, fill in required information, such as personally identifiable information,
Occupational information and contact information etc. improve authentication, recognition of face and vivo identification, bind phone number and bank card
Deng, for promoted customer experience the skills such as optical character identification (Optical Character Recognition, OCR) also can be used
Art obtains data, it is therefore intended that the authenticity for ensuring basic information prevents puppet from emitting application.
Step S20 is searched pair according to the identity label information in loan application request in the big data platform
The loan transaction information answered.
It is understood that the correctness in order to guarantee the Transaction Information obtained, the loan transaction information of user is passed through
Identity label information is managed, according to the corresponding loan transaction information of the identity label acquisition of information of user, the identity mark
The unique identification that information is user is signed, such as ID=100 etc. can also be used as label information, the present embodiment pair by other means
This is with no restriction.
Step S30 extracts the current transaction feature information in the loan transaction information.
In the present embodiment, the loan transaction information can be compared with preset keyword information, by the loan
Contain the target transaction information of the preset keyword information in Transaction Information as the current transaction feature information, wherein
The preset keyword information can be the keyword message with the refund probability correlation of user such as the amount of the loan and refund date,
To carry out the extraction of effective information by keyword message, the accuracy of the target refund probabilistic forecasting of user is improved.
Step S40, obtain preset rules information, according to the preset rules information to the current transaction feature information into
Row extension, obtains target transaction characteristic information.
It should be noted that the preset rules information is derivative for polynomial characteristic, using the derivative engineering of polynomial characteristic
Technology, the data sources such as integrated use Transaction Information, account information and customer information, from exchange hour, transaction amount, trading floor
Institute, trading frequency and transaction spend speed etc. to carry out Multi-Dimensional Extension, to obtain prediction data more in all directions.
In the present embodiment, feature derivative is that existing feature is combined, and the new feature for entering mould is formed, after derivative
Feature have to target signature more tremendous influence and classifying quality, to improve the prediction effect of model.
Step S50 carries out refund probability by the default credit prediction model according to the target transaction characteristic information
Prediction, obtains the target refund probability results of the user.
In the present embodiment, the default credit prediction model may be based on Taxonomy and distribution (Classification
And Regression Trees, CART) generating algorithm constructed, default credit prediction model is generated using CART tree,
It is predicted by default credit prediction model, thus more precisely identification positioning fraudulent trading.
The flow diagram of an embodiment as shown in Figure 3, in Classical forecast, to pass through the request for data for obtaining user
When, it can be applied first into part, then by admittable regulation, select the client of access according to different product characteristic, such as age,
Occupation, the admittable regulations such as region select target visitor group, then pass through a variety of anti-fraudulent mean identification fraud visitors by anti-fraud rule
Family, and by credit rating, customer data is collected by authorization, develops credit scoring model, gives a mark, can be applied to client
Risk Pricing, client refuse rule, calculate client's income by the accrediting amount and are in debt, comprehensive descision client's loan repayment capacity,
The accrediting amount is formulated to client, it is finally decided whether the client promoter that makes loans draws application, fund directly gets to client's binding
In bank card.
It is derivative by the derivative feature that carries out of polynomial characteristic first in this embodiment, feature selecting is then carried out, and in mould
It selects CART classification tree generation algorithm to be predicted in type platform, decides whether to make loans according to prediction result.
The present embodiment through the above scheme, is requested by obtaining the loan application that user initiates in big data platform,
In, the loan application request includes identity label information;According to the identity label information in loan application request in institute
It states and searches corresponding loan transaction information in big data platform;Extract the current transaction feature letter in the loan transaction information
Breath;Preset rules information is obtained, the current transaction feature information is extended according to the preset rules information, obtains mesh
Mark transaction feature information;It is pre- that refund probability carried out by the default credit prediction model according to the target transaction characteristic information
It surveys, the target refund probability results of the user is obtained, thus according to the preset rules information to the current transaction feature
Information is extended, and obtains more fully predicted characteristics information, and carry out refund probabilistic forecasting by default credit prediction model,
To improve the accuracy for prediction of refunding.
In one embodiment, as shown in figure 4, proposing that the present invention is based on the refund of big data platform is pre- based on first embodiment
Survey method second embodiment, before the step S10, the method also includes:
Obtain the connection status with the big data platform.
In the present embodiment, by carrying out data analysis with big data platform, so as to be obtained from big data platform
The customer transaction data of more various dimensions.
When the connection status is to connect normal, step S10 is executed.
In one embodiment, the preset rules information includes default exhibition opening information, the step S40, comprising:
Step S401 obtains default exhibition opening information, according to the default exhibition opening information to the current transaction feature
Information is extended, and obtains target transaction characteristic information.
Further, the step S401, comprising:
Default exhibition opening information is obtained, it is special to the current transaction by following formula according to the default exhibition opening information
Reference breath is extended, and obtains target transaction characteristic information;
The formula are as follows:
Wherein, index a, b, c, m are full
The combination of sufficient a+b+c+m≤n, and i≤j≤k≤l);
The J indicates that target transaction characteristic information, the m indicate current dimension, and n indicates default exhibition opening information, a, b, c
And m indicates that the index information of default item, x indicate extension feature information, i, j, k and l respectively indicate target transaction feature letter
Cease corresponding item sequence number.
In tool is realized, the default expansion information is the degree of polynomial expansion, is indicated with degree, belongs to Int type,
When carrying out the expansion that degree is n to m feature, the feature of note m dimension is { x1,x2,...xm, then expansion is m shared, often
Item is as follows:
1st:
2nd:Wherein index a, b are all combinations (wherein i≤j) for meeting a+b≤n;
3rd:Wherein index a, b, c be meet a+b+c≤n all combinations (wherein i≤j≤
k);
...
M: be the sum of m-1 first:(its
Middle i≤j≤k).
Such as: input feature vector are as follows: { x, y }, and when degree is 2, derivative feature are as follows: x, x*x, y, x*y, y*y then have altogether
Derive three features.
Characteristic information is carried out derivative expansion by multinomial, thus by current feature by scheme provided in this embodiment
Information expansion is the characteristic information of various dimensions, when carrying out user in predicting, reaches the accuracy for improving user in predicting.
In one embodiment, as shown in figure 5, proposing that the present invention is based on big datas based on the first embodiment or the second embodiment
The refund prediction technique 3rd embodiment of platform is illustrated based on first embodiment in the present embodiment, the step S50
Before, the method also includes:
Obtain historical trading characteristic information and corresponding history target refund probabilistic information;The historical trading feature is believed
Composition characteristic data set is ceased, and using the characteristic data set as node;The default history that the characteristic is concentrated is extracted to hand over
Easy characteristic information, and the default corresponding history target refund probabilistic information of historical trading characteristic information is obtained, according to the history
The characteristic data set is divided into the first data set and the second data set by target refund probabilistic information;If the training dataset of node
For D, calculates existing feature and the gini index of the data set is obtained according to first data set and the second data set with reference to base
Buddhist nun's index.
At this point, to each feature A, each value a that it may be taken, according to sample point to the test of A=a be "Yes" or
D is divided into D by "No"1And D2Two parts calculate gini index using following formula;
From described with reference to the corresponding historical trading characteristic information of selection preset reference gini index and history in gini index
Target refund probabilistic information is as child node;
It should be noted that the preset reference gini index is the smallest historical trading characteristic information of gini index.
When the child node meets preset condition, the default credit is obtained according to the node and child node and is predicted
Model.
In the present embodiment, in all possible feature A and the possible cut-off a of all of which, Geordie is selected to refer to
The smallest feature of number and its corresponding cut-off are as optimal characteristics and optimal cut-off.According to optimal characteristics and optimal cut-off,
Two child nodes are generated from existing node, training dataset is assigned in two child nodes according to feature, two child nodes are passed
It calls with returning, until meeting stop condition, to generate default credit prediction model, wherein the preset condition is in node
The number of historical trading characteristic information be less than the gini index of predetermined threshold or historical trading characteristic information and be less than predetermined threshold
Value or historical trading characteristic information have been trained.
In one embodiment, described when the child node meets preset condition, it is obtained according to the node and child node
To the default credit prediction model, comprising:
When the child node meets preset condition, reference credit is generated according to the node and child node and predicts mould
Type;Sample transaction feature information is obtained, reference credit prediction model described in the sample transaction feature information input is gone back
Money probabilistic forecasting obtains prediction refund probability results.
In the present embodiment, in order to improve the accuracy of prediction, the reference credit prediction model trained can be tested
Card, goes out the corresponding model of the higher prediction result of precision of prediction as default credit prediction model for verification result.
In the concrete realization, by obtaining the corresponding sample refund probability results of sample transaction feature information, by the sample
This refund probability results are compared with the prediction refund probability results, are obtained objective appraisal index, are commented according to the target
Valence index selects target credit prediction model from the reference credit prediction model, and the target credit prediction model is made
For the default credit prediction model.
In one embodiment, after the step S50, the method also includes:
The target refund probability results are compared by step S501 with preset threshold.
It should be noted that the preset threshold be 70%, can also be other parameters, the present embodiment to this with no restriction,
In the present embodiment, it is illustrated for 70%, such as when the target refund probability results predicted are 50%, with 70%
It is compared, is then less than preset threshold, in this case, target refund probability results are small with preset threshold, then illustrate user
Refund probability it is little.
Step S502 is obtained in the loan application request when the target refund probability results are more than preset threshold
User's shroff account number information, according to loan application described in user's shroff account number information response request.
In the present embodiment, in order to improve loan efficiency, when the target refund probability results of user are more than preset threshold,
Then illustrate that user is top-tier customer, then dozen money directly can be carried out by the shroff account number information of user, facilitate user can carry out and
When provide a loan, improve user experience.
Scheme provided in this embodiment verifies the reference credit prediction model trained by sample data, will
The refund probability results that reference credit prediction model predicts are compared with actual refund probability results, by precision of prediction compared with
The corresponding reference credit prediction model of high prediction result is as default credit prediction model, to improve the precision of model.
The refund prediction meanss based on big data platform that the present invention further provides a kind of.
It is that the present invention is based on the functional modules of the refund prediction meanss first embodiment of big data platform to show referring to Fig. 6, Fig. 6
It is intended to.
The present invention is based in the refund prediction meanss first embodiment of big data platform, it is somebody's turn to do the refund based on big data platform
Prediction meanss include:
Module 10 is obtained, the loan application request initiated in big data platform for obtaining user, wherein the loan
Application request includes identity label information.
In the present embodiment, risk of fraud information data system, including thousands of category informations have been built based on big data technology
Content and nearly 10,000,000,000 transaction records cover client-related information, account information, card image, authorization message, clearance letter
Breath, refund information and have been acknowledged transaction swindling record etc. data, to obtain the Transaction Information of user.
In the concrete realization, apply for user Ke Tong big data platform, fill in required information, such as personally identifiable information,
Occupational information and contact information etc. improve authentication, recognition of face and vivo identification, bind phone number and bank card
Deng, for promoted customer experience the skills such as optical character identification (Optical Character Recognition, OCR) also can be used
Art obtains data, it is therefore intended that the authenticity for ensuring basic information prevents puppet from emitting application.
Searching module 20, for the identity label information in being requested according to the loan application in the big data platform
Search corresponding loan transaction information.
It is understood that the correctness in order to guarantee the Transaction Information obtained, the loan transaction information of user is passed through
Identity label information is managed, according to the corresponding loan transaction information of the identity label acquisition of information of user, the identity mark
The unique identification that information is user is signed, such as ID=100 etc. can also be used as label information, the present embodiment pair by other means
This is with no restriction.
Extraction module 30, for extracting the current transaction feature information in the loan transaction information.
In the present embodiment, the loan transaction information can be compared with preset keyword information, by the loan
Contain the target transaction information of the preset keyword information in Transaction Information as the current transaction feature information, wherein
The preset keyword information can be the keyword message with the refund probability correlation of user such as the amount of the loan and refund date,
To carry out the extraction of effective information by keyword message, the accuracy of the target refund probabilistic forecasting of user is improved.
Expansion module 40, it is special to the current transaction according to the preset rules information for obtaining preset rules information
Reference breath is extended, and obtains target transaction characteristic information.
It should be noted that the preset rules information is derivative for polynomial characteristic, using the derivative engineering of polynomial characteristic
Technology, the data sources such as integrated use Transaction Information, account information and customer information, from exchange hour, transaction amount, trading floor
Institute, trading frequency and transaction spend speed etc. to carry out Multi-Dimensional Extension, to obtain prediction data more in all directions.
In the present embodiment, feature derivative is that existing feature is combined, and the new feature for entering mould is formed, after derivative
Feature have to target signature more tremendous influence and classifying quality, to improve the prediction effect of model.
Prediction module 50, for being gone back according to the target transaction characteristic information by the default credit prediction model
Money probabilistic forecasting obtains the target refund probability results of the user.
In the present embodiment, the default credit prediction model may be based on Taxonomy and distribution (Classification
And Regression Trees, CART) generating algorithm constructed, default credit prediction model is generated using CART tree,
It is predicted by default credit prediction model, thus more precisely identification positioning fraudulent trading.
The flow diagram of an embodiment as shown in Figure 3, in Classical forecast, to pass through the request for data for obtaining user
When, it can be applied first into part, then by admittable regulation, select the client of access according to different product characteristic, such as age,
Occupation, the admittable regulations such as region select target visitor group, then pass through a variety of anti-fraudulent mean identification fraud visitors by anti-fraud rule
Family, and by credit rating, customer data is collected by authorization, develops credit scoring model, gives a mark, can be applied to client
Risk Pricing, client refuse rule, calculate client's income by the accrediting amount and are in debt, comprehensive descision client's loan repayment capacity,
The accrediting amount is formulated to client, it is finally decided whether the client promoter that makes loans draws application, fund directly gets to client's binding
In bank card.
It is derivative by the derivative feature that carries out of polynomial characteristic first in this embodiment, feature selecting is then carried out, and in mould
It selects CART classification tree generation algorithm to be predicted in type platform, decides whether to make loans according to prediction result.
The present embodiment through the above scheme, is requested by obtaining the loan application that user initiates in big data platform,
In, the loan application request includes identity label information;According to the identity label information in loan application request in institute
It states and searches corresponding loan transaction information in big data platform;Extract the current transaction feature letter in the loan transaction information
Breath;Preset rules information is obtained, the current transaction feature information is extended according to the preset rules information, obtains mesh
Mark transaction feature information;It is pre- that refund probability carried out by the default credit prediction model according to the target transaction characteristic information
It surveys, the target refund probability results of the user is obtained, thus according to the preset rules information to the current transaction feature
Information is extended, and obtains more fully predicted characteristics information, and carry out refund probabilistic forecasting by default credit prediction model,
To improve the accuracy for prediction of refunding.
In addition, the embodiment of the present invention also proposes a kind of storage medium, it is stored on the storage medium flat based on big data
The refund Prediction program of platform, the refund Prediction program based on big data platform be executed by processor it is as described above based on
The step of refund prediction technique of big data platform.
It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row
His property includes, so that the process, method, article or the device that include a series of elements not only include those elements, and
And further include other elements that are not explicitly listed, or further include for this process, method, article or device institute it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including being somebody's turn to do
There is also other identical elements in the process, method of element, article or device.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but in many cases
The former is more preferably embodiment.Based on this understanding, technical solution of the present invention substantially in other words does the prior art
The part contributed out can be embodied in the form of software products, which is stored in one as described above
In computer readable storage medium (such as ROM/RAM, magnetic disk, CD), including some instructions are with so that an intelligent terminal is set
Standby (can be mobile phone, computer, terminal device, air conditioner or network-termination device etc.) executes each embodiment of the present invention
The method.
The above is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (10)
1. a kind of refund prediction technique based on big data platform, which is characterized in that the refund based on big data platform is pre-
Survey method includes:
Obtain the loan application request that user initiates in big data platform, wherein the loan application request includes identity mark
Sign information;
Corresponding loan transaction is searched in the big data platform according to the identity label information in loan application request
Information;
Extract the current transaction feature information in the loan transaction information;
Preset rules information is obtained, the current transaction feature information is extended according to the preset rules information, is obtained
Target transaction characteristic information;
Refund probabilistic forecasting is carried out by the default credit prediction model according to the target transaction characteristic information, is obtained described
The target refund probability results of user.
2. the refund prediction technique based on big data platform as described in claim 1, which is characterized in that the acquisition user exists
Before the loan application request initiated in big data platform, the method also includes:
Obtain the connection status with the big data platform;
When the connection status is to connect normal, executes the loan application that the acquisition user initiates in big data platform and ask
The step of asking.
3. the refund prediction technique based on big data platform as described in claim 1, which is characterized in that the preset rules letter
Breath includes default exhibition opening information;
The acquisition preset rules information is extended the current transaction feature information according to the preset rules information,
Obtain target transaction characteristic information, comprising:
Default exhibition opening information is obtained, the current transaction feature information is extended according to the default exhibition opening information,
Obtain target transaction characteristic information.
4. the refund prediction technique based on big data platform as claimed in claim 3, which is characterized in that obtain default expansion degree
Information is extended the current transaction feature information by following formula according to the default exhibition opening information, obtains mesh
Mark transaction feature information;
The formula are as follows:
Wherein, index a, b, c, m are to meet a+b
+ c+m≤n combination, and i≤j≤k≤l);
Wherein, the J indicates that target transaction characteristic information, the m indicate current dimension, and n expression, which is preset, opens up opening information, a, b,
C and m indicates that the index information of default item, x indicate extension feature information, and i, j, k and l respectively indicate target transaction feature letter
Cease corresponding item sequence number.
5. the refund prediction technique based on big data platform according to any one of claims 1 to 4, which is characterized in that institute
It states and refund probabilistic forecasting is carried out by the default credit prediction model according to the target transaction characteristic information, obtain the use
Before the target refund probability results at family, the method also includes:
Obtain historical trading characteristic information and corresponding history target refund probabilistic information;
By the historical trading characteristic information composition characteristic data set, and using the characteristic data set as node;
The default historical trading characteristic information that the characteristic is concentrated is extracted, and it is corresponding to obtain default historical trading characteristic information
History target refund probabilistic information, according to the history target refund probabilistic information by the characteristic data set be divided into first number
According to collection and the second data set;
It is obtained according to first data set and the second data set with reference to gini index;
From described with reference to the corresponding historical trading characteristic information of selection preset reference gini index and history target in gini index
Refund probabilistic information is as child node;
When the child node meets preset condition, the default credit is obtained according to the node and child node and predicts mould
Type.
6. the refund prediction technique based on big data platform as claimed in claim 5, which is characterized in that described in the sub- section
When point meets preset condition, the default credit prediction model is obtained according to the node and child node, comprising:
When the child node meets preset condition, reference credit prediction model is generated according to the node and child node;
Sample transaction feature information is obtained, reference credit prediction model described in the sample transaction feature information input is gone back
Money probabilistic forecasting obtains prediction refund probability results;
Obtain the corresponding sample refund probability results of sample transaction feature information, by the sample refund probability results with it is described pre-
It surveys refund probability results to be compared, obtains objective appraisal index;
Target credit prediction model is selected from the reference credit prediction model according to the objective appraisal index, it will be described
Target credit prediction model is as the default credit prediction model.
7. the refund prediction technique based on big data platform according to any one of claims 1 to 4, which is characterized in that institute
Stating loan application request includes user's shroff account number information;
It is described that refund probabilistic forecasting is carried out by the default credit prediction model according to the target transaction characteristic information, it obtains
After the target refund probability results of the user, the method also includes:
The target refund probability results are compared with preset threshold;
When the target refund probability results are more than preset threshold, user's shroff account number in the loan application request is obtained
Information is requested according to loan application described in user's shroff account number information response.
8. a kind of refund prediction meanss based on big data platform, which is characterized in that the refund based on big data platform is pre-
Surveying device includes:
Module is obtained, the loan application request initiated in big data platform for obtaining user, wherein the loan application is asked
It asks including identity label information;
Searching module, for being searched in the big data platform pair according to the identity label information in loan application request
The loan transaction information answered;
Extraction module, for extracting the current transaction feature information in the loan transaction information;
Expansion module, for obtaining preset rules information, according to the preset rules information to the current transaction feature information
It is extended, obtains target transaction characteristic information;
Prediction module, for carrying out refund probability by the default credit prediction model according to the target transaction characteristic information
Prediction, obtains the target refund probability results of the user.
9. a kind of pre- measurement equipment of refund based on big data platform, which is characterized in that the refund based on big data platform is pre-
Measurement equipment include: memory, processor and be stored on the memory and can run on the processor based on big number
According to the refund Prediction program of platform, the refund Prediction program based on big data platform is arranged for carrying out such as claim 1 to 7
Any one of described in the refund prediction technique based on big data platform the step of.
10. a kind of storage medium, which is characterized in that be stored with the pre- ranging of refund based on big data platform on the storage medium
Sequence is realized when the refund Prediction program based on big data platform is executed by processor such as any one of claims 1 to 7 institute
The step of refund prediction technique based on big data platform stated.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910342059.1A CN110276677A (en) | 2019-04-24 | 2019-04-24 | Refund prediction technique, device, equipment and storage medium based on big data platform |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910342059.1A CN110276677A (en) | 2019-04-24 | 2019-04-24 | Refund prediction technique, device, equipment and storage medium based on big data platform |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110276677A true CN110276677A (en) | 2019-09-24 |
Family
ID=67959592
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910342059.1A Pending CN110276677A (en) | 2019-04-24 | 2019-04-24 | Refund prediction technique, device, equipment and storage medium based on big data platform |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110276677A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
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 |
CN111079991A (en) * | 2019-11-29 | 2020-04-28 | 珠海随变科技有限公司 | Service index prediction method, device, equipment and storage medium |
CN111369342A (en) * | 2020-03-05 | 2020-07-03 | 中国建设银行股份有限公司 | Loan approval method, device, equipment and storage medium based on machine learning |
CN111913994A (en) * | 2020-08-12 | 2020-11-10 | 武汉众邦银行股份有限公司 | Client risk data monitoring method based on inline data and external data |
CN112991052A (en) * | 2021-04-25 | 2021-06-18 | 大箴(杭州)科技有限公司 | Repayment capability evaluation method and device |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8244618B1 (en) * | 2009-01-08 | 2012-08-14 | Amherst Holdings, LLC | Loan information analysis system and method |
CN108256691A (en) * | 2018-02-08 | 2018-07-06 | 成都智宝大数据科技有限公司 | Refund Probabilistic Prediction Model construction method and device |
CN108428188A (en) * | 2018-01-24 | 2018-08-21 | 顺丰科技有限公司 | Claims Resolution Risk Forecast Method, system, equipment and storage medium |
CN108846520A (en) * | 2018-06-22 | 2018-11-20 | 北京京东金融科技控股有限公司 | Overdue loan prediction technique, device and computer readable storage medium |
CN109034658A (en) * | 2018-08-22 | 2018-12-18 | 重庆邮电大学 | A kind of promise breaking consumer's risk prediction technique based on big data finance |
CN109472683A (en) * | 2017-09-08 | 2019-03-15 | 平安普惠企业管理有限公司 | Customer lending qualification generation method, server and readable storage medium storing program for executing |
-
2019
- 2019-04-24 CN CN201910342059.1A patent/CN110276677A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8244618B1 (en) * | 2009-01-08 | 2012-08-14 | Amherst Holdings, LLC | Loan information analysis system and method |
CN109472683A (en) * | 2017-09-08 | 2019-03-15 | 平安普惠企业管理有限公司 | Customer lending qualification generation method, server and readable storage medium storing program for executing |
CN108428188A (en) * | 2018-01-24 | 2018-08-21 | 顺丰科技有限公司 | Claims Resolution Risk Forecast Method, system, equipment and storage medium |
CN108256691A (en) * | 2018-02-08 | 2018-07-06 | 成都智宝大数据科技有限公司 | Refund Probabilistic Prediction Model construction method and device |
CN108846520A (en) * | 2018-06-22 | 2018-11-20 | 北京京东金融科技控股有限公司 | Overdue loan prediction technique, device and computer readable storage medium |
CN109034658A (en) * | 2018-08-22 | 2018-12-18 | 重庆邮电大学 | A kind of promise breaking consumer's risk prediction technique based on big data finance |
Non-Patent Citations (1)
Title |
---|
裔隽等: "《Python机器学习实战》", 28 February 2018, 科学技术文献出版社 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
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 |
CN111079991A (en) * | 2019-11-29 | 2020-04-28 | 珠海随变科技有限公司 | Service index prediction method, device, equipment and storage medium |
CN111369342A (en) * | 2020-03-05 | 2020-07-03 | 中国建设银行股份有限公司 | Loan approval method, device, equipment and storage medium based on machine learning |
CN111369342B (en) * | 2020-03-05 | 2023-10-17 | 中国建设银行股份有限公司 | Loan approval method, device, equipment and storage medium based on machine learning |
CN111913994A (en) * | 2020-08-12 | 2020-11-10 | 武汉众邦银行股份有限公司 | Client risk data monitoring method based on inline data and external data |
CN111913994B (en) * | 2020-08-12 | 2023-09-15 | 武汉众邦银行股份有限公司 | Customer risk data monitoring method based on in-line data and external data |
CN112991052A (en) * | 2021-04-25 | 2021-06-18 | 大箴(杭州)科技有限公司 | Repayment capability evaluation method and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110276677A (en) | Refund prediction technique, device, equipment and storage medium based on big data platform | |
US7328218B2 (en) | Constrained tree structure method and system | |
CN108256691A (en) | Refund Probabilistic Prediction Model construction method and device | |
CN108876600A (en) | Warning information method for pushing, device, computer equipment and medium | |
CN108648074A (en) | Loan valuation method, apparatus based on support vector machines and equipment | |
CN109035003A (en) | Anti- fraud model modelling approach and anti-fraud monitoring method based on machine learning | |
CN108492104A (en) | A kind of resource transfer monitoring method and device | |
CN109583904A (en) | Training method, impaired operation detection method and the device of abnormal operation detection model | |
US20210264448A1 (en) | Privacy preserving ai derived simulated world | |
CN107689008A (en) | A kind of user insures the method and device of behavior prediction | |
CN108921686A (en) | A kind of credit-graded approach and device of personal user | |
CN110288350A (en) | User's Value Prediction Methods, device, equipment and storage medium | |
CN110288459A (en) | Loan prediction technique, device, equipment and storage medium | |
CN110083765A (en) | Financing information-pushing method, device, computer equipment and storage medium | |
CN109325845A (en) | A kind of financial product intelligent recommendation method and system | |
CN110288460A (en) | Collection prediction technique, device, equipment and storage medium based on propagated forward | |
CN110276369A (en) | Feature selection approach, device, equipment and storage medium based on machine learning | |
CN109615280A (en) | Employee's data processing method, device, computer equipment and storage medium | |
CN108960912A (en) | Method and apparatus for determining target position | |
CN109800363A (en) | Construct method, apparatus, equipment and the storage medium of standing relational network | |
CN110349013A (en) | Risk control method and device | |
CN110728301A (en) | Credit scoring method, device, terminal and storage medium for individual user | |
CN110197426A (en) | A kind of method for building up of credit scoring model, device and readable storage medium storing program for executing | |
CN109829593A (en) | The credit rating of target object determines method, apparatus, storage medium and electronic device | |
Nagaraju et al. | Methodologies used for customer churn detection in customer relationship management |
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: 20190924 |