CN109767071A - User credit ranking method, device, computer equipment and storage medium - Google Patents
User credit ranking method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN109767071A CN109767071A CN201811535575.8A CN201811535575A CN109767071A CN 109767071 A CN109767071 A CN 109767071A CN 201811535575 A CN201811535575 A CN 201811535575A CN 109767071 A CN109767071 A CN 109767071A
- Authority
- CN
- China
- Prior art keywords
- model
- xgboost
- parameter
- credit
- initial
- 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
Abstract
This application involves the disaggregated model fields of artificial intelligence, and in particular to a kind of user credit ranking method, device, computer equipment and storage medium.Method included: the credit rating data obtained to Ratings User；And the XGBoost-BYS credit classification model for obtaining the input of the credit rating data to Ratings User, the credit classification results to Ratings User are obtained by the model.And the establishment process of model includes: to obtain initial XGBoost model and model initial parameter, is initialized according to model initial parameter to initial XGBoost model；And the initial parameter of initial XGBoost model is optimized by Bayes's optimization, obtain XGBoost-BYS credit classification model.The application optimizes model parameter by Bayes's optimization, the parameter dependence of XGBoost model can effectively be got rid of, cumbersome, time-consuming and the randomness, unstability that can be avoided artificial parameter adjustment are conducive to more efficient Ratings User for the treatment of and carry out credit rating.
Description
Technical field
This application involves field of computer technology, set more particularly to a kind of user credit ranking method, device, computer
Standby and storage medium.
Background technique
Personal credit is the basis of entire society's credit.Main market players is made of individual, all in marketing
Economic activity, it is closely bound up with personal credit.Once the constraint of personal behavior mistake will occur personal discreditable behavior, and then go out
Existing collective breaks one's promise.Therefore, individual credit system construction is extremely important.Personal credit is not only a national market
The basis of ethics and Moral Culture construction, the huge resource of even more one national economic development.It opens up and utilizes this resource, energy
Consumption is effectively pushed, optimizes allocation of resources, promotes economic development.Market economy is developed more, and the function that personal credit is played is got over
It is important, individual credit system it is perfect whether have become market economy whether one of Cheng Shu distinctive marks.
Personal credit file under big data background more and more uses the data such as network behavior, consumption, social activity, such
Data are different from traditional reference information, so that traditional individual credit risk assessment models and method can not obtain promising result.
Current random forest, GBDT, XGBoost scheduling algorithm have apparent classification performance advantage, but there is parameters
The disadvantage of dependence is also difficult to reach ideal sort effect using the traditional combination of the hyper parameter such as grid search evolutionary algorithm.
Summary of the invention
Based on this, it is necessary in view of the above technical problems, provide a kind of model evaluation method that parameter dependence is low, dress
It sets, computer equipment and storage medium.
A kind of user credit ranking method, which comprises
Obtain the credit rating data to Ratings User；
The credit rating data to Ratings User are inputted into XGBoost-BYS credit classification model, obtain it is described to
The credit classification results of Ratings User；
The acquisition process of the XGBoost-BYS credit classification model includes: to obtain initial XGBoost model and model
Initial parameter initializes the initial XGBoost model according to the model initial parameter, the initial XGBoost
Model, which is used to treat Ratings User according to credit rating data, carries out credit classification；
The initial parameter of initial XGBoost model after optimizing the parameter initialization by Bayes, obtains
XGBoost-BYS credit classification model.
In one of the embodiments, it is described establish initial XGBoost model and model initial parameter be set specifically include:
The initial XGBoost model and model initial parameter using classification accuracy as fitness function are obtained, according to
The model initial parameter initializes the initial XGBoost model.
The initial ginseng for optimizing the initial XGBoost model by Bayes in one of the embodiments,
Number obtains XGBoost-BYS credit classification model and specifically includes:
Initial XGBoost model is obtained, according to Bayesian Optimization Algorithm to the default initial of the initial XGBoost model
Parameter carries out parameter adjustment；
Obtain the corresponding multiple groups value value of the initial XGBoost model after the adjustment of every subparameter；
The corresponding parameter value of highest value value in the multiple groups value value is obtained, highest value value is corresponding
Parameter value is as targeted parameter value；
XGBoost Model Parameter is set according to the targeted parameter value, obtains XGBoost-BYS credit classification model.
It is described in one of the embodiments, that XGBoost Model Parameter is arranged according to the targeted parameter value, it obtains
XGBoost-BYS credit classification model specifically includes:
Classified by the XGBoost model after setting parameter to the sample data of tape label, records the setting ginseng
The classification accuracy of XGBoost model after number；
The parameter value in XGBoost model after adjusting the setting parameter according to the classification accuracy, by adjusting
XGBoost model after parameter classifies to the sample data of tape label, the XGBoost model after recording the setting parameter
Classification accuracy；
The change rate for obtaining the classification accuracy will be newest when the change rate of the accuracy rate is lower than threshold value
XGBoost model is as XGBoost-BYS credit classification model.
It is described in one of the embodiments, that XGBoost Model Parameter is arranged according to the targeted parameter value, it obtains
XGBoost-BYS credit classification model specifically includes:
Classified by the XGBoost model after setting parameter to the sample data of tape label, records the setting ginseng
The classification accuracy of XGBoost model after number；
The parameter value in XGBoost model after adjusting the setting parameter according to the classification accuracy, by adjusting
XGBoost model after parameter classifies to the sample data of tape label, and the number of parameter adjustment；
When parameter adjustment number is greater than adjustment frequency threshold value, using newest XGBoost model as XGBoost-
BYS credit classification model.
The credit rating data to Ratings User are inputted into the XGBoost-BYS in one of the embodiments,
Credit classification model obtains the credit classification results to Ratings User and specifically includes:
The credit rating data to Ratings User are inputted into the XGBoost-BYS credit classification model；
Classified according to the decision tree in the XGBoost-BYS credit classification model to the credit rating data,
Each decision tree is obtained to the preliminary classification result of the credit rating data classification；
The credit classification results to Ratings User are obtained according to the preliminary classification result.
A kind of user credit grading device, described device include:
Rating information obtains module, for obtaining the credit rating data to Ratings User；
Credit rating module, for the credit rating data to Ratings User to be inputted XGBoost-BYS credit score
Class model obtains the credit classification results to Ratings User；
The acquisition process of the XGBoost-BYS credit classification model includes: to obtain initial XGBoost model and model
Initial parameter initializes the initial XGBoost model according to the model initial parameter, the initial XGBoost
Model, which is used to treat Ratings User according to credit rating data, carries out credit classification；
The initial parameter of initial XGBoost model after optimizing the parameter initialization by Bayes, obtains
XGBoost-BYS credit classification model.
The model optimization module specifically includes in one of the embodiments:
Parameter adjustment unit, for obtaining initial XGBoost model, according to Bayesian Optimization Algorithm to described initial
The default initial parameter of XGBoost model carries out parameter adjustment；
Evaluation unit obtains the corresponding multiple groups value value of the initial XGBoost model after the adjustment of every subparameter；
Target component acquiring unit obtains the corresponding parameter value of highest value value in the multiple groups value value, will most
The high corresponding parameter value of value value is as targeted parameter value；
Model optimization unit is arranged XGBoost Model Parameter according to the targeted parameter value, obtains XGBoost-BYS
Credit classification model.
A kind of computer equipment, including memory and processor, the memory are stored with computer program, the processing
Device performs the steps of when executing the computer program
Obtain the credit rating data to Ratings User；
The credit rating data to Ratings User are inputted into XGBoost-BYS credit classification model, obtain it is described to
The credit classification results of Ratings User；
The acquisition process of the XGBoost-BYS credit classification model includes: to obtain initial XGBoost model and model
Initial parameter initializes the initial XGBoost model according to the model initial parameter, the initial XGBoost
Model, which is used to treat Ratings User according to credit rating data, carries out credit classification；At the beginning of optimizing the parameter by Bayes
The initial parameter of initial XGBoost model after beginningization obtains XGBoost-BYS credit classification model.
A kind of computer readable storage medium, is stored thereon with computer program, and the computer program is held by processor
It is performed the steps of when row
Obtain the credit rating data to Ratings User；
The credit rating data to Ratings User are inputted into XGBoost-BYS credit classification model, obtain it is described to
The credit classification results of Ratings User；
The acquisition process of the XGBoost-BYS credit classification model includes: to obtain initial XGBoost model and model
Initial parameter initializes the initial XGBoost model according to the model initial parameter, the initial XGBoost
Model, which is used to treat Ratings User according to credit rating data, carries out credit classification；At the beginning of optimizing the parameter by Bayes
The initial parameter of initial XGBoost model after beginningization obtains XGBoost-BYS credit classification model.
Above-mentioned user credit ranking method, device, computer equipment and storage medium, by obtaining to Ratings User
Credit rating data；And the XGBoost-BYS credit classification model for obtaining the input of the credit rating data to Ratings User, lead to
Cross credit classification results of the model acquisition to Ratings User.And the establishment process of model includes: to obtain initial XGBoost model
And model initial parameter, the initial XGBoost model is initialized according to the model initial parameter；And pass through shellfish
Ye Si optimization optimizes the initial parameter of initial XGBoost model, obtains XGBoost-BYS credit classification model.This Shen
XGBoost model is established based on XGBoost algorithm in scheme please, but XGBoost algorithm has stronger parameter dependence, institute
To be optimized by Bayes's optimization to XGBoost model, XGBoost-BYS credit hierarchy model is obtained, Bayes is passed through
Optimization can effectively get rid of the parameter dependence of XGBoost model, can be avoided the cumbersome, time-consuming of artificial parameter adjustment with
Machine, unstability are conducive to more efficient Ratings User for the treatment of and carry out credit rating wait the individual that grades.
Detailed description of the invention
Fig. 1 is the applied environment figure of user credit ranking method in one embodiment；
Fig. 2 is the flow diagram of user credit ranking method in one embodiment；
Fig. 3 is the flow diagram of user credit ranking method in one embodiment；
The sub-step flow diagram that Fig. 4 is step S400 in Fig. 2 in one embodiment；
Fig. 5 is the flow diagram of user credit ranking method in one embodiment；
Fig. 6 is the structural block diagram of user credit grading device in one embodiment；
Fig. 7 is the internal structure chart of computer equipment in one embodiment.
Specific embodiment
It is with reference to the accompanying drawings and embodiments, right in order to which the objects, technical solutions and advantages of the application are more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, not
For limiting the application.
User credit ranking method provided by the present application, can be applied in application environment as shown in Figure 1, wherein letter
It is communicated by network with server with the terminal 102 where classification staff, server 104 obtains credit by network
The credit rating data to Ratings User that staff provides are classified, after server receives credit rating data, based on pre-
If XGboost-BYS credit hierarchy model, by credit hierarchy model be based on credit rating data treat Ratings User carry out
Classification results are then fed back to terminal 102 by credit classification, and terminal 102 can be, but not limited to be various personal computers, notes
This computer, smart phone, tablet computer.
As shown in Fig. 2, the model evaluation method of the application in one of the embodiments, can be realized by server,
Specifically includes the following steps:
S600 obtains the credit rating data to Ratings User.
It refers to waiting the client being rated personal to Ratings User, grading, which can according to need, to be arranged to Ratings User
Credit grade.Credit rating data refer to the data graded for treating Ratings User, can specifically include: individual borrows
Refund record, credit card purchase record and the data such as individual subscriber work and family data.
After establishment can be used for the model of credit rating, so that it may treat the credit of Ratings User using the model
It grades.Server is obtained first to the relevant credit rating data of Ratings User.
Credit rating data to Ratings User are inputted XGBoost-BYS credit classification model, obtained wait grade by S800
The credit classification results of user.
Credit classification results refer to the credit grade of individual subscriber to be graded, which, which can according to need, is set as more
A grade, different brackets division represent the credit rating different to Ratings User.
The XGBoost-BYS credit classification model that the credit rating data input of acquisition is obtained, by passing through engineering
The credit that the XGBoost-BYS credit classification model that parameter adjusts after habit treats Ratings User is graded.
It is primarily based on to treat the classification accuracy foundation that Ratings User carries out credit classification according to credit rating data
Initial XGBoost model, is then adjusted by parameter of the Bayesian Optimization Algorithm to initial XGBoost model, then again
The secondary model according to after optimization treats Ratings User and carries out the classification accuracy of credit classification to be adjusted to parameter, by not
Disconnected adjustment carrys out the classification accuracy of lift scheme, obtains final XGBoost-BYS credit classification model.
The establishment process of XGBoost-BYS credit classification model specifically includes:
S200 obtains initial XGBoost model and model initial parameter, according to model initial parameter to initial
XGBoost model is initialized, and initial XGBoost model, which is used to treat Ratings User according to credit rating data, carries out credit
Classification.To treat classification accuracy that Ratings User is classified as the model fitness function of initial XGBoost model.
XGBoost model is a kind of machine learning model established based on XGBoost algorithm.XGBoost algorithm is a kind of
One of Gradient boosting algorithm, common Gradient boosting algorithm is represented as GBDT (Gradient
Boosting Decision Tree, gradient promote decision tree).And XGBoost algorithm is Gradient boosting algorithm
A kind of efficient way of realization.Be initialized as a constant first, Gradient boosting be according to first derivative ri,
XGBoost is to be added renewal learning device according to first derivative gi and second dervative hi, grey iterative generation base learner.Compare tradition
Gradient boosting algorithm, XGBoost has following feature, it is contemplated that the case where training data is sparse value, can
Think the default direction of missing values or specified value assigned finger, this can greatly promote the efficiency of algorithm；After characteristic series sequence
It is stored in memory in the form of block, be may be reused in iteration；Although boosting algorithm iteration must be serial,
It can accomplish when handling each characteristic series parallel；It considers when data volume is bigger, how when Out of Memory effectively uses
Disk, can by combine multithreading, data compression, fragment method, as far as possible improve algorithm efficiency.Model is initially joined
Number refers to parameters in initial model, in this application, min_child_weight that initial parameter mainly includes,
The parameters such as Cosample_bytree, max_depth, Subsample, Eta, Gamma.Min_child_weight is to determine most
Small leaf node sample weights and.It, can be to avoid model learning to the special sample of part when its value is larger.But if this
A value is excessively high, will lead to poor fitting.Cosample_bytree is used to control the accounting of the columns of every tree stochastical sampling.max_
Depth is used to control the depth capacity of tree.Subsample is for controlling for each tree, the ratio of stochastical sampling.Reduce this
The value-based algorithm of parameter can be guarded more, and over-fitting is avoided.But the setting of this value is too small, it may result in poor fitting.Eta
Robustness for improved model.Least disadvantage function drop-out value needed for Gamma specifies node split.This parameter value is got over
Greatly, algorithm is more conservative.Individual to be graded is referred to Ratings User.
S400 optimizes the initial parameter of initial XGBoost model by Bayes, obtains XGBoost-BYS credit
Disaggregated model.In one of the embodiments, to treat the classification success rate of Ratings User as the mould of initial XGBoost model
Type fitness function.
Bayes's optimization is a kind of method of close approximation.If we do not know that some function is specifically, that
It is what that some known priori knowledges, which may just be will use, and approach or guess the function.This is just exactly the core of posterior probability
Thought.Assuming that there is a series of observation samples, and data are one and connect a ground investment model and be trained.In this way after training
Model will obey significantly some function, and the unknown function also will depend entirely on the data that it is acquired.Due to not knowing
What the model fitness function for best effects of classifying in XGBoost-BYS credit classification model is, it is possible to pass through pattra leaves
This optimization is sought come the carry out approximation of the function best to classifying quality.The model fitness function of initial XGBoost model
Parameter is set as initial value, by Bayes optimize method in XGBoost model min_child_weight,
The parameters such as Cosample_bytree, max_depth, Subsample, Eta, Gamma carry out tune ginseng, i.e. model fitness function
In parameters be adjusted, improve the classification accuracy of XGBoost model, can then obtain be more suitable and be used for
Accurately treat the XGBoost-BYS credit classification model that Ratings User carries out credit classification.Establishing initial XGBoost model
Afterwards, the parameter of initial XGBoost model is adjusted by Bayes's optimization, acquisition can treat Ratings User and accurately be divided
The XGBoost-BYS credit classification model of class.
Above-mentioned user credit ranking method, by obtaining the credit rating data to Ratings User；And it will be to Ratings User
The XGBoost-BYS credit classification model that obtains of credit rating data input, letter to Ratings User is obtained by the model
Use classification results.And the establishment process of model includes: to obtain initial XGBoost model and model initial parameter, according to model
Initial parameter initializes initial XGBoost model；And by Bayes's optimization to the initial ginseng of initial XGBoost model
Number optimizes, and obtains XGBoost-BYS credit classification model.XGBoost model is calculated based on XGBoost in the scheme of the application
Method is established, but XGBoost algorithm has stronger parameter dependence, so being carried out by Bayes's optimization to XGBoost model
Optimization obtains XGBoost-BYS credit hierarchy model, and the parameter of XGBoost model can be effectively got rid of by Bayes's optimization
Dependence can be avoided cumbersome, time-consuming and the randomness, unstability of artificial parameter adjustment, be conducive to more efficient treat
Ratings User is i.e. wait personal progress credit rating of grading.
As shown in figure 3, S200 is specifically included in one of the embodiments:
S210 obtains initial XGBoost model and model initial parameter using classification accuracy as fitness function,
Initial XGBoost model is initialized according to model initial parameter.
It is available to treat the accuracy rate of the credit rating of Ratings User as the fitness letter of initial XGBoost model
Several initial XGBoost models, and initial XGBoost model is initialized based on model initial parameter, by with credit
The accuracy rate of grading can effectively establish the credit rating for treating Ratings User as the fitness function of beginning XGBoost model
XGBoost model.
As shown in figure 4, S400 is specifically included in one of the embodiments:
S410 obtains initial XGBoost model, according to Bayesian Optimization Algorithm to the pre- of the initial XGBoost model
If initial parameter carries out parameter adjustment；
S430 obtains the corresponding multiple groups value value of initial XGBoost model after the adjustment of every subparameter.
S450 obtains the corresponding parameter value of highest value value in multiple groups value value, and highest value value is corresponding
Parameter value is as targeted parameter value；
S470 is arranged XGBoost Model Parameter according to targeted parameter value, obtains XGBoost-BYS credit classification model.
It is the optimum results of Bayes's optimization that value value is corresponding, and value value is bigger, illustrates that the result of optimization is better.It is first
Bayes's optimization is first passed through to be adjusted initial parameter default in initial XGBoost model.Parameter preset can specifically include
The parameters such as min_child_weight, Cosample_bytree, max_depth, Subsample, Eta, Gamma and alpha
Deng.And it obtains every subparameter parameter combination adjusted and corresponds to value value, and choose maximum corresponding one group of wherein value value
Parameter combination, and according to this group of parameter sets to min_child_weight, Cosample_ in initial XGBoost model
The parameters such as bytree, max_depth, Subsample, Eta, Gamma and alpha are configured, and obtain XGBoost-BYS letter
Use disaggregated model.
S470 is specifically included in one of the embodiments:
Classified by the XGBoost model after setting parameter to the sample data of tape label, after record setting parameter
XGBoost model classification accuracy.The parameter in XGBoost model after adjusting setting parameter according to classification accuracy
Value, classifies to the sample data of tape label by adjusting the XGBoost model after parameter, after record setting parameter
The classification accuracy of XGBoost model.The change rate for obtaining classification accuracy will when the change rate of accuracy rate is lower than threshold value
Newest XGBoost model is as XGBoost-BYS credit classification model.
The maximum corresponding one group of parameter of value value that Bayes optimizes acquisition is carried out to initial XGBoost model obtaining
After combination, first the parameter in initial XGBoost model can be configured according to this group of parameter combination, then pass through setting ginseng
Model after number classifies to the sample data of tape label, obtains the accuracy rate of classification, while again based on the result of the classification
Parameter in XGBoost model currently in use is finely adjusted, and again by the XGBoost model after fine tuning to tape label
Sample data classify.When the change rate of the accuracy rate of classification is lower than preset threshold value, stop adjusting parameter,
And be adjusted according to parameter of the parameter being finally arranged to initial XGBoost model, obtain XGBoost-BYS credit classification mould
Type.Parameter is adjusted repeatedly by the training result based on parameter model adjusted, is conducive to the classification for improving model
Accuracy rate.
S470 is specifically included in one of the embodiments:
Classified by the XGBoost model after setting parameter to the sample data of tape label, after record setting parameter
XGBoost model classification accuracy；The parameter in XGBoost model after adjusting setting parameter according to classification accuracy
Value, and the number of recording parameters adjustment；When the number of parameter adjustment is greater than frequency threshold value, newest XGBoost model is made
For XGBoost-BYS credit classification model.
The maximum corresponding one group of parameter of value value that Bayes optimizes acquisition is carried out to initial XGBoost model obtaining
After combination, first the parameter in initial XGBoost model can be configured according to this group of parameter combination, then pass through setting ginseng
Model after number classifies to the sample data of tape label, obtains the accuracy rate of classification, while again based on the result of the classification
Parameter in XGBoost model currently in use is finely adjusted, and again by the XGBoost model after fine tuning to tape label
Sample data classify.Until recycle time, that is, adjusting parameter numerical value number has been more than preset adjustment frequency threshold value
When, stop adjusting parameter, and be adjusted according to parameter of the parameter being finally arranged to initial XGBoost model, obtains
XGBoost-BYS credit classification model.Parameter is adjusted repeatedly by the training result based on parameter model adjusted,
Be conducive to improve the classification accuracy of model.
S800 is specifically included in one of the embodiments: the credit rating data to Ratings User are inputted
XGBoost-BYS credit classification model；According to the decision tree in XGBoost-BYS credit classification model to credit rating data into
Row classification, obtains each decision tree to the preliminary classification result of credit rating data classification；According to preliminary classification result obtain to
The credit classification results of Ratings User.
It is treated first by each decision tree of XGBoost-BYS credit classification model according to the credit rating data of acquisition
The credit rating of Ratings User is classified, and the classification of all decision trees in XGBoost-BYS credit classification model is then integrated
As a result, the credit grade classification of Ratings User is treated, it in one of the embodiments, can be by decision trees most in classification results
Obtained classification results are as the credit classification results for treating Ratings User.
As shown in figure 5, S800 in one of the embodiments, later further include:
S900 is generated according to credit classification results and is reported to the personal credit file of Ratings User.
After the credit rating of Ratings User is treated in acquisition, a credit can be generated based on the result of the credit rating
Grading report, the particular content of report may include the rating result and some record either promise breaking notes of keeping one's word of user
Record.By credit rating report can the credit situation more clearly to user summarize, show simultaneously.
The user credit ranking method of the application is specifically includes the following steps: obtain to divide in one of the embodiments,
Initial XGBoost model and model initial parameter of the class accuracy rate as fitness function, according to model initial parameter to first
Beginning XGBoost model is initialized；By Bayes's optimization to the min_child_weight of initial XGBoost model,
Cosample_bytree, max_depth, Subsample, Eta, Gamma and alpha carry out parameter adjustment；Obtain ginseng every time
The corresponding multiple groups value value of initial XGBoost model after number adjustment；It is corresponding to obtain highest value value in multiple groups value value
Parameter value, using the corresponding parameter value of highest value value as targeted parameter value；Pass through the XGBoost model after setting parameter
Classify to the sample data of tape label, the classification accuracy of the XGBoost model after record setting parameter；It is quasi- according to classification
The parameter value in XGBoost model after true rate adjustment setting parameter, by adjusting the XGBoost model after parameter to tape label
Sample data classify,；The change rate for obtaining classification accuracy will be newest when the change rate of accuracy rate is lower than threshold value
XGBoost model as XGBoost-BYS credit classification model.Obtain the credit rating data to Ratings User；It will be to be evaluated
The credit rating data of grade user input XGBoost-BYS credit classification model, obtain the credit classification results to Ratings User.
And it is generated according to credit classification results and is reported to the personal credit file of Ratings User.
It should be understood that although each step in the flow chart of Fig. 2-5 is successively shown according to the instruction of arrow,
These steps are not that the inevitable sequence according to arrow instruction successively executes.Unless expressly stating otherwise herein, these steps
Execution there is no stringent sequences to limit, these steps can execute in other order.Moreover, at least one in Fig. 2-5
Part steps may include that perhaps these sub-steps of multiple stages or stage are not necessarily in synchronization to multiple sub-steps
Completion is executed, but can be executed at different times, the execution sequence in these sub-steps or stage is also not necessarily successively
It carries out, but can be at least part of the sub-step or stage of other steps or other steps in turn or alternately
It executes.
The device as shown in fig. 6, a kind of user credit is graded, device include:
Rating information obtains module 200, for obtaining the credit rating data to Ratings User；
Credit rating module 400, for the credit rating data to Ratings User to be inputted XGBoost-BYS credit classification
Model obtains the credit classification results to Ratings User；
The acquisition process of XGBoost-BYS credit classification model includes: that the initial XGBoost model of acquisition and model are initial
Parameter initializes initial XGBoost model according to model initial parameter, and initial XGBoost model is used for according to credit
Ratings data treats Ratings User and carries out credit classification；
The initial parameter for optimizing the initial XGBoost model after Optimal Parameters initialization by Bayes, obtains
XGBoost-BYS credit classification model.
In one of the embodiments, further include model building module, for using classification accuracy as it is to be created just
The fitness function of beginning XGBoost model establishes initial XGBoost model and model initial parameter is arranged.
Model optimization module 400 specifically includes in one of the embodiments: parameter adjustment unit, initial for obtaining
XGBoost model carries out parameter tune according to default initial parameter of the Bayesian Optimization Algorithm to the initial XGBoost model
It is whole；；Evaluation unit obtains the corresponding multiple groups value value of initial XGBoost model after the adjustment of every subparameter；Target component obtains
Unit obtains the corresponding parameter value of highest value value in multiple groups value value, by the corresponding parameter value work of highest value value
For targeted parameter value；Model optimization unit is arranged XGBoost Model Parameter according to targeted parameter value, obtains XGBoost-BYS
Credit classification model.
Model optimization unit is specifically used in one of the embodiments: passing through the XGBoost model pair after setting parameter
The sample data of tape label is classified, the classification accuracy of the XGBoost model after record setting parameter；It is accurate according to classification
The parameter value in XGBoost model after rate adjustment setting parameter, by adjusting the XGBoost model after parameter to tape label
Sample data is classified, the classification accuracy of the XGBoost model after recording adjusting parameter；Obtain the variation of classification accuracy
Rate, when the change rate of accuracy rate is lower than threshold value, using newest XGBoost model as XGBoost-BYS credit classification model.
Model optimization unit is specifically used in one of the embodiments: passing through the XGBoost model pair after setting parameter
The sample data of tape label is classified, the classification accuracy of the XGBoost model after record setting parameter；It is accurate according to classification
The parameter value in XGBoost model after rate adjustment setting parameter, by adjusting the XGBoost model after parameter to tape label
The step of sample data is classified, the classification accuracy of the XGBoost model after recording adjusting parameter, and adjusting parameter time
Number；When adjusting parameter number is greater than frequency threshold value, using newest XGBoost model as XGBoost-BYS credit classification mould
Type.
In one embodiment, credit rating module 400 is specifically used for: the credit rating data to Ratings User are inputted
XGBoost-BYS credit classification model；According to the decision tree in XGBoost-BYS credit classification model to credit rating data into
Row classification, obtains each decision tree to the preliminary classification result of credit rating data classification；According to preliminary classification result obtain to
The credit classification results of Ratings User.
It in one of the embodiments, further include that report generation module is used to generate use to be graded according to credit classification results
The personal credit file at family is reported.
Specific about user credit grading device limits the limit that may refer to above for user credit ranking method
Fixed, details are not described herein.Modules in above-mentioned user credit grading device can fully or partially through software, hardware and its
Combination is to realize.Above-mentioned each module can be embedded in the form of hardware or independently of in the processor in computer equipment, can also be with
It is stored in the memory in computer equipment in a software form, in order to which processor calls the above modules of execution corresponding
Operation.
In one embodiment, a kind of computer equipment is provided, which can be server, internal junction
Composition can be as shown in Figure 7.The computer equipment include by system bus connect processor, memory, network interface and
Database.Wherein, the processor of the computer equipment is for providing calculating and control ability.The memory packet of the computer equipment
Include non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system, computer program and data
Library.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating
The database of machine equipment is for storing user credit ratings data.The network interface of the computer equipment is used for and external terminal
It is communicated by network connection.To realize a kind of user credit ranking method when the computer program is executed by processor.
It will be understood by those skilled in the art that structure shown in Fig. 7, only part relevant to application scheme is tied
The block diagram of structure does not constitute the restriction for the computer equipment being applied thereon to application scheme, specific computer equipment
It may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
In one embodiment, a kind of computer equipment, including memory and processor are provided, is stored in memory
Computer program, the processor perform the steps of when executing computer program
Obtain the credit rating data to Ratings User；
Credit rating data to Ratings User are inputted into XGBoost-BYS credit classification model, are obtained to Ratings User
Credit classification results；
The acquisition process of XGBoost-BYS credit classification model includes: that the initial XGBoost model of acquisition and model are initial
Parameter initializes initial XGBoost model according to model initial parameter, and initial XGBoost model is used for according to credit
Ratings data treats Ratings User and carries out credit classification；
The initial parameter for optimizing the initial XGBoost model after Optimal Parameters initialization by Bayes, obtains
XGBoost-BYS credit classification model.
In one embodiment, acquisition is performed the steps of when processor executes computer program also with classification accuracy
As the initial XGBoost model and model initial parameter of fitness function, according to model initial parameter to initial XGBoost
Model is initialized.
In one embodiment, it is also performed the steps of when processor executes computer program
Initial XGBoost model is obtained, according to Bayesian Optimization Algorithm to the default initial of the initial XGBoost model
Parameter carries out parameter adjustment；Obtain the corresponding multiple groups value value of initial XGBoost model after the adjustment of every subparameter；Obtain multiple groups
The corresponding parameter value of highest value value in value value, using the corresponding parameter value of highest value value as targeted parameter value；
XGBoost Model Parameter is set according to targeted parameter value, obtains XGBoost-BYS credit classification model.
In one embodiment, it is also performed the steps of when processor executes computer program by after setting parameter
XGBoost model classifies to the sample data of tape label, and the classification of the XGBoost model after record setting parameter is accurate
Rate；The parameter value in XGBoost model after adjusting setting parameter according to classification accuracy, after parameter
XGBoost model classifies to the sample data of tape label, and the classification of the XGBoost model after record setting parameter is accurate
Rate；Obtain classification accuracy change rate, when the change rate of accuracy rate be lower than threshold value when, using newest XGBoost model as
XGBoost-BYS credit classification model.
In one embodiment, it is also performed the steps of when processor executes computer program by after setting parameter
XGBoost model classifies to the sample data of tape label, and the classification of the XGBoost model after record setting parameter is accurate
Rate；The parameter value in XGBoost model after adjusting setting parameter according to classification accuracy, after parameter
XGBoost model classifies to the sample data of tape label, and the number of parameter adjustment；When parameter adjustment number is greater than adjustment
When frequency threshold value, using newest XGBoost model as XGBoost-BYS credit classification model.
In one embodiment, it also performs the steps of when processor executes computer program by the letter to Ratings User
XGBoost-BYS credit classification model is inputted with ratings data；According to the decision tree pair in XGBoost-BYS credit classification model
Credit rating data are classified, and obtain each decision tree to the preliminary classification result of credit rating data classification；According to primary
Classification results obtain the credit classification results to Ratings User.
In one embodiment, it also performs the steps of when processor executes computer program according to credit classification results
It generates and is reported to the personal credit file of Ratings User.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program performs the steps of when being executed by processor
Obtain the credit rating data to Ratings User；
Credit rating data to Ratings User are inputted into XGBoost-BYS credit classification model, are obtained to Ratings User
Credit classification results；
The acquisition process of XGBoost-BYS credit classification model includes: that the initial XGBoost model of acquisition and model are initial
Parameter initializes initial XGBoost model according to model initial parameter, and initial XGBoost model is used for according to credit
Ratings data treats Ratings User and carries out credit classification；
The initial parameter for optimizing the initial XGBoost model after Optimal Parameters initialization by Bayes, obtains
XGBoost-BYS credit classification model.
In one embodiment, it is accurate to classify that acquisition is also performed the steps of when computer program is executed by processor
Initial XGBoost model and model initial parameter of the rate as fitness function, according to model initial parameter to initial
XGBoost model is initialized.
In one embodiment, it is also performed the steps of when computer program is executed by processor and obtains initial XGBoost
Model carries out parameter adjustment according to default initial parameter of the Bayesian Optimization Algorithm to the initial XGBoost model；It obtains every
The corresponding multiple groups value value of initial XGBoost model after subparameter adjustment；Obtain highest value value pair in multiple groups value value
The parameter value answered, using the corresponding parameter value of highest value value as targeted parameter value；It is arranged according to targeted parameter value
XGBoost Model Parameter obtains XGBoost-BYS credit classification model.
In one embodiment, it is also performed the steps of when computer program is executed by processor by after setting parameter
XGBoost model classify to the sample data of tape label, record setting parameter after XGBoost model classification it is accurate
Rate；The parameter value in XGBoost model after adjusting setting parameter according to classification accuracy, after parameter
XGBoost model classifies to the sample data of tape label, and the classification of the XGBoost model after record setting parameter is accurate
Rate；Obtain classification accuracy change rate, when the change rate of accuracy rate be lower than threshold value when, using newest XGBoost model as
XGBoost-BYS credit classification model.
In one embodiment, it is also performed the steps of when computer program is executed by processor by after setting parameter
XGBoost model classify to the sample data of tape label, record setting parameter after XGBoost model classification it is accurate
Rate；The parameter value in XGBoost model after adjusting setting parameter according to classification accuracy, after parameter
XGBoost model classifies to the sample data of tape label, and the number of parameter adjustment；When parameter adjustment number is greater than adjustment
When frequency threshold value, using newest XGBoost model as XGBoost-BYS credit classification model.
In one embodiment, also performing the steps of when computer program is executed by processor will be to Ratings User
Credit rating data input XGBoost-BYS credit classification model；According to the decision tree in XGBoost-BYS credit classification model
Classify to credit rating data, obtains each decision tree to the preliminary classification result of credit rating data classification；According to first
Grade classification results obtain the credit classification results to Ratings User.
In one embodiment, it also performs the steps of to be classified according to credit when computer program is executed by processor and tie
Fruit generates reports to the personal credit file of Ratings User.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Instruct relevant hardware to complete by computer program, computer program to can be stored in a non-volatile computer readable
It takes in storage medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, this Shen
Please provided by any reference used in each embodiment to memory, storage, database or other media, may each comprise
Non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield all should be considered as described in this specification.
Above embodiments only express the several embodiments of the application, and the description thereof is more specific and detailed, but can not
Therefore it is construed as limiting the scope of the patent.It should be pointed out that for those of ordinary skill in the art,
Under the premise of not departing from the application design, various modifications and improvements can be made, these belong to the protection scope of the application.
Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (10)
1. a kind of user credit ranking method, which comprises
Obtain the credit rating data to Ratings User；
The credit rating data to Ratings User are inputted into XGBoost-BYS credit classification model, are obtained described wait grade
The credit classification results of user；
The acquisition process of the XGBoost-BYS credit classification model includes: that the initial XGBoost model of acquisition and model are initial
Parameter initializes the initial XGBoost model according to the model initial parameter, the initial XGBoost model
Credit classification is carried out for treating Ratings User according to credit rating data, the parameter initialization is optimized by Bayes
The initial parameter of initial XGBoost model afterwards obtains XGBoost-BYS credit classification model.
2. the method according to claim 1, wherein the initial XGBoost model of the acquisition and model are initial
Parameter, carrying out initialization to the initial XGBoost model according to the model initial parameter includes:
The initial XGBoost model and model initial parameter using classification accuracy as fitness function are obtained, according to described
Model initial parameter initializes the initial XGBoost model.
3. the method according to claim 1, wherein described described initial by Bayes's optimization optimization
The initial parameter of XGBoost model obtains XGBoost-BYS credit classification model and specifically includes:
Initial XGBoost model is obtained, according to Bayesian Optimization Algorithm to the default initial parameter of the initial XGBoost model
Carry out parameter adjustment；
Obtain the corresponding multiple groups value value of the initial XGBoost model after the adjustment of every subparameter；
The corresponding parameter value of highest value value in the multiple groups value value is obtained, by the corresponding parameter of highest value value
Value is used as targeted parameter value；
XGBoost Model Parameter is set according to the targeted parameter value, obtains XGBoost-BYS credit classification model.
4. according to the method described in claim 3, it is characterized in that, described be arranged XGBoost mould according to the targeted parameter value
Parameter in type obtains XGBoost-BYS credit classification model and specifically includes:
Classified by the XGBoost model after setting parameter to the sample data of tape label, after recording the setting parameter
XGBoost model classification accuracy；
The parameter value in XGBoost model after adjusting the setting parameter according to the classification accuracy, by adjusting parameter
XGBoost model afterwards classifies to the sample data of tape label, point of the XGBoost model after recording the setting parameter
Class accuracy rate；
The change rate for obtaining the classification accuracy, when the change rate be lower than threshold value when, using newest XGBoost model as
XGBoost-BYS credit classification model.
5. according to the method described in claim 3, it is characterized in that, described be arranged XGBoost mould according to the targeted parameter value
Parameter in type obtains XGBoost-BYS credit classification model and specifically includes:
Classified by the XGBoost model after setting parameter to the sample data of tape label, after recording the setting parameter
XGBoost model classification accuracy；
The parameter value in XGBoost model after adjusting the setting parameter according to the classification accuracy, by adjusting parameter
XGBoost model afterwards classifies to the sample data of tape label, and the number of recording parameters adjustment；
When parameter adjustment number is greater than adjustment frequency threshold value, believe newest XGBoost model as XGBoost-BYS
Use disaggregated model.
6. the method according to claim 1, wherein described that the credit rating data to Ratings User are defeated
Enter the XGBoost-BYS credit classification model, obtain the credit classification results to Ratings User and specifically include:
The credit rating data to Ratings User are inputted into the XGBoost-BYS credit classification model；
Classified according to the decision tree in the XGBoost-BYS credit classification model to the credit rating data, is obtained
Preliminary classification result of each decision tree to the credit rating data classification；
The credit classification results to Ratings User are obtained according to the preliminary classification result.
The device 7. a kind of user credit is graded, which is characterized in that described device includes:
Rating information obtains module, for obtaining the credit rating data to Ratings User；
Credit rating module, for the credit rating data to Ratings User to be inputted XGBoost-BYS credit classification mould
Type obtains the credit classification results to Ratings User；
The acquisition process of the XGBoost-BYS credit classification model includes: that the initial XGBoost model of acquisition and model are initial
Parameter initializes the initial XGBoost model according to the model initial parameter, the initial XGBoost model
Credit classification is carried out for treating Ratings User according to credit rating data；
The initial parameter of initial XGBoost model after optimizing the parameter initialization by Bayes, obtains
XGBoost-BYS credit classification model.
8. device according to claim 7, which is characterized in that it further include model optimization module, the model optimization module
It specifically includes:
Parameter adjustment unit, for obtaining initial XGBoost model, according to Bayesian Optimization Algorithm to the initial XGBoost
The default initial parameter of model carries out parameter adjustment；
Evaluation unit obtains the corresponding multiple groups value value of the initial XGBoost model after the adjustment of every subparameter；
Target component acquiring unit obtains the corresponding parameter value of highest value value in the multiple groups value value, will be highest
The corresponding parameter value of value value is as targeted parameter value；
Model optimization unit is arranged XGBoost Model Parameter according to the targeted parameter value, obtains XGBoost-BYS credit
Disaggregated model.
9. a kind of computer equipment, including memory and processor, the memory are stored with computer program, feature exists
In, the processor realized when executing the computer program claim 1 to 6 to any one of the method the step of.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program
The step of method described in any one of claims 1 to 6 is realized when being executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811535575.8A CN109767071A (en) | 2018-12-14 | 2018-12-14 | User credit ranking method, device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811535575.8A CN109767071A (en) | 2018-12-14 | 2018-12-14 | User credit ranking method, device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109767071A true CN109767071A (en) | 2019-05-17 |
Family
ID=66451874
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811535575.8A Pending CN109767071A (en) | 2018-12-14 | 2018-12-14 | User credit ranking method, device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109767071A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110598995A (en) * | 2019-08-15 | 2019-12-20 | 中国平安人寿保险股份有限公司 | Intelligent customer rating method and device and computer readable storage medium |
| CN111950585A (en) * | 2020-06-29 | 2020-11-17 | 广东技术师范大学 | XGboost-based underground comprehensive pipe gallery safety condition assessment method |
| CN113536746A (en) * | 2021-07-26 | 2021-10-22 | 平安资产管理有限责任公司 | Credit rating report generation method and device, computer equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107733886A (en) * | 2017-09-30 | 2018-02-23 | 天津大学 | The application layer ddos attack detection method that a kind of logic-based returns |
| CN108154430A (en) * | 2017-12-28 | 2018-06-12 | 上海氪信信息技术有限公司 | A kind of credit scoring construction method based on machine learning and big data technology |
| US20180322406A1 (en) * | 2017-05-04 | 2018-11-08 | Zestfinance, Inc. | Systems and methods for providing machine learning model explainability information |
-
2018
- 2018-12-14 CN CN201811535575.8A patent/CN109767071A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180322406A1 (en) * | 2017-05-04 | 2018-11-08 | Zestfinance, Inc. | Systems and methods for providing machine learning model explainability information |
| CN107733886A (en) * | 2017-09-30 | 2018-02-23 | 天津大学 | The application layer ddos attack detection method that a kind of logic-based returns |
| CN108154430A (en) * | 2017-12-28 | 2018-06-12 | 上海氪信信息技术有限公司 | A kind of credit scoring construction method based on machine learning and big data technology |
Non-Patent Citations (1)
| Title |
|---|
| 金欣: "个人信用评估指标及模型研究——基于XGBoost-BOA集成分类模型", 《中国优秀硕士学位论文全文数据库经济与管理科学辑》, pages 39 - 40 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110598995A (en) * | 2019-08-15 | 2019-12-20 | 中国平安人寿保险股份有限公司 | Intelligent customer rating method and device and computer readable storage medium |
| CN110598995B (en) * | 2019-08-15 | 2023-08-25 | 中国平安人寿保险股份有限公司 | Smart client rating method, smart client rating device and computer readable storage medium |
| CN111950585A (en) * | 2020-06-29 | 2020-11-17 | 广东技术师范大学 | XGboost-based underground comprehensive pipe gallery safety condition assessment method |
| CN113536746A (en) * | 2021-07-26 | 2021-10-22 | 平安资产管理有限责任公司 | Credit rating report generation method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110175120B (en) | Fuzzy test case self-adaptive mutation method and device based on reinforcement learning | |
| Voss et al. | TMVA, the toolkit for multivariate data analysis with ROOT | |
| CN109492026B (en) | Telecommunication fraud classification detection method based on improved active learning technology | |
| CN109829153A (en) | Intension recognizing method, device, equipment and medium based on convolutional neural networks | |
| CN109543032A (en) | File classification method, device, computer equipment and storage medium | |
| CN109376995A (en) | Financial data methods of marking, device, computer equipment and storage medium | |
| CN109767071A (en) | User credit ranking method, device, computer equipment and storage medium | |
| CN111950656B (en) | Image recognition model generation method and device, computer equipment and storage medium | |
| CN109886275A (en) | Reproduction image-recognizing method, device, computer equipment and storage medium | |
| Chellapandi et al. | Comparison of pre-trained models using transfer learning for detecting plant disease | |
| CN110826618A (en) | Personal credit risk assessment method based on random forest | |
| CN111738522B (en) | Photovoltaic power generation power prediction method, storage medium and terminal equipment | |
| WO2023116111A1 (en) | Disk fault prediction method and apparatus | |
| CN109104731A (en) | Construction method, device and the computer equipment of cell scenario category classification model | |
| CN109214904A (en) | Acquisition methods, device, computer equipment and the storage medium of financial fraud clue | |
| Gao et al. | A mobile application for plant recognition through deep learning | |
| Jindal et al. | Offline handwritten Gurumukhi character recognition system using deep learning | |
| CN115563610A (en) | Method and device for training and identifying intrusion detection model | |
| Gu et al. | Leaf species recognition based on VGG16 networks and transfer learning | |
| CN106951728A (en) | A kind of tumour key gene recognition methods based on particle group optimizing and marking criterion | |
| Li et al. | Utilizing the LightGBM Algorithm for Operator User Credit Assessment Research | |
| CN109977977A (en) | A kind of method and corresponding intrument identifying potential user | |
| CN109754135A (en) | Behavior of credit data processing method, device, storage medium and computer equipment | |
| Kuchipudi et al. | Android Malware Detection using Ensemble Learning | |
| Irawan et al. | Accounts Receivable Seamless Prediction for Companies by Using Multiclass Data Mining Model |
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 |