CN107330781A - A kind of individual credit risk appraisal procedure based on IFOA SVM - Google Patents
A kind of individual credit risk appraisal procedure based on IFOA SVM Download PDFInfo
- Publication number
- CN107330781A CN107330781A CN201710463836.9A CN201710463836A CN107330781A CN 107330781 A CN107330781 A CN 107330781A CN 201710463836 A CN201710463836 A CN 201710463836A CN 107330781 A CN107330781 A CN 107330781A
- Authority
- CN
- China
- Prior art keywords
- svm
- drosophila
- credit
- ifoa
- algorithm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Finance (AREA)
- General Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of individual credit risk appraisal procedure based on IFOA SVM, by being improved to traditional drosophila algorithm, effectively ensure the low optimization accuracy of algorithm while algorithm Searching efficiency can be improved.IFOA is to find out preferably n elite drosophila position of current flavor concentration, is weighted processing to these positions, obtains an optimal weighting position, allows drosophila group to be flown to the position.Let operation fly away to carry out n time, select that flavor concentration average is optimal once lets fly away.Improved drosophila algorithm not only increases the probability for finding optimal solution, also accelerated the speed for finding optimal solution, applied in assessing credit risks, obtain preferable assessment result, and by the quantification treatment to credit scoring model in sample data, estimated performance is improved, with stronger practicality, for the clear and definite loan customer credit of bank and other financial mechanism, reduction credit risk provides effective foundation.
Description
Technical field
The present invention relates to assessing credit risks method, more particularly to based on the personal credit for improving drosophila algorithm optimization SVM
Methods of risk assessment, belongs to utilization of the artificial intelligence approach in assessing credit risks field.
Background technology
Without mortgaging under the situation that pure fiduciary loan heat constantly heats up, each business bank assign loan transaction as development
Emphasis.But Banking Supervision Commission's issue in 2016《China's banking industry operation report》The non-performing loan remaining sum of five big rows is 8079.5 hundred million
Member, non-performing loan rate increases by 25.7% on a year-on-year basis.The main cause for hindering credit operation development is business bank to credit risk
Managerial skills are relatively low, lack effective personal credit file method.The accuracy rate of Credit Risk Assessment Model often rises one hundred
Branch, so that it may the profit in terms of ten thousand is brought for business bank.Therefore research individual credit risk is assessed with very strong practical valency
Value.
Artificial intelligence model can preferably solve this non-linear similar to assessing credit risks compared to statistical model
Pattern classification problem.Conventional artificial intelligence model have Bayesian network, decision tree (Decision Trees, DT), support to
Amount machine (Support Vector Machine, SVM) and BP neural network etc..SVM is especially solving small-scale sample, non-thread
During the problems such as property and high dimensional pattern are recognized, good performance can be shown, therefore SVM is also obtained in assessing credit risks
It is widely applied.But numerous studies show：SVM performances are preferable, but can not always obtain best effect.SVM has very strong
Learning ability and generalization ability, but SVM model prediction performances quality and parameter selection it is closely related.Therefore using has
The method of effect searches optimal SVM parameters, and it is the hot issue studied at present to obtain the higher classification degree of accuracy.Domestic and foreign scholars
The method of relevant supporting vector machine model parameter (the penalty factor and kernel functional parameter) optimization proposed mainly includes gradient and declined
Method (Gradient Descent, GD), particle swarm optimization algorithm (Particle Swarm Optimization, PSO), drosophila
Optimized algorithm (FOA) and genetic algorithm (Genetic Algorithm, GA) etc..Such as Jiang Minghui (2007) is to overcome artificial choosing
The randomness of parameter is selected, with genetic algorithm optimization SVM parameters, precision of prediction is improved.Wang, et al (2013) proposition fruit
Fly algorithm is optimized to SVM parameters, applies to the prediction of ship operation, precision of prediction is than PSO-SVM and GA-SVM prediction essences
Degree is all high.For traditional drosophila optimized algorithm when to parameter optimization, the defect of local extremum is easily trapped into.The present invention is to FOA
Improved, it is proposed that a kind of improved drosophila optimized algorithm optimizes SVM parameter, and be applied to credit risk and commented
In estimating.In order to embody the superiority of the model evaluation effect, with commenting for SVM, GA-SVM and FOA-SVM of gridding method search parameter
Estimate effect to be contrasted, the results show IFOA-SVM models can obtain higher accuracy rate in assessing credit risks.
The content of the invention
It is an object of the invention to when to parameter optimization, be easily trapped into local extremum for traditional drosophila optimized algorithm
Defect, is improved there is provided a kind of individual credit risk appraisal procedure for improving drosophila algorithm optimization SVM to drosophila algorithm, has
Effect improves the accuracy rate of assessing credit risks.
The present invention provides a kind of individual credit risk appraisal procedure based on IFOA-SVM, including：
Step 1), credit scoring model is quantified according to the personal data of creditor, it is determined that based on IFOA-SVM
The object function of people's credit risk assessment model；
Step 2), it is used for carrying out creditor the SVM models of credit evaluation based on SVMs foundation, discriminant function isWherein K (xi, yj) it is kernel function, b is constant, aiFor Lagrange factor i=1,2...n；
Step 3), drosophila algorithm is improved.
Step 4), using improved drosophila algorithm optimization SVM parameters, with improved drosophila algorithm to penalty factor and core
Function g carries out global optimizing, obtains the optimal solution of two parameters.Optimal solution parameter is updated to the SVM models in step 2
Training is practised, the individual credit risk assessment models based on IFOA-SVM are set up；
Step 5), by the part in step 1 be used as test creditor's data be brought into step 4 based on IFOA-SVM
Individual credit risk assessment models, and carried out with SVM, GA-SVM and FOA-SVM of gridding method search parameter Evaluated effect
Contrast, the comparing result of four kinds of models is as shown in table 1.
The class model classification results of table 1 four are contrasted
The step 1 includes following sub-step：
(1,1) personal credit's data used in the present invention are the 5th, source " safe enlightening cup " data mining challenge matches.Choosing
1000 borrower's information and its corresponding credit rating information are taken, wherein 500 groups as training data, 500 groups are used to test
Data.Training sample set and test sample collection distribution situation are as shown in table 2：
The sample set distribution situation of table 2
(1,2) choose the state of bank account, continue month, credit history, credit amount degree, savings account account
Family/bond, current working condition, marital relations, personal occupancy, real estate, the age, instalment plan, in this family silver
The existing amount of credit of row, available guarantor, whether telephone number whether there is register, as this 15 standard diagrams of foreign personage
As input variable, what is finally exported is its credit rating：1 represents risky, and 0 represents devoid of risk.
The index of table 3 is selected and quantization method
(1,3) object function is set to the root-mean-square error after SVM is predicted to data, and causes mean square error minimum
Change.
The step 2 includes following sub-step：
(2,1) for given linear separability data setIt is linear in d dimension spaces
Discriminant function g (x)=wx+b, then can use hyperplane:It is normal vector that wx+b=0, which carries out w in sample separation, formula, and b is inclined
Move.In the case of linear separability, using yi(wxi+b) -1 >=0, i=1,2...n is represented.
(2,2), can be by slack variable ξ when linearly inseparableiIt is incorporated into constraints,It is middle to introduce penalty factor to solve the problem.
(2,3) introduce kernel function when being nonlinear problemWherein σ ＞ 0；
(2,4) the SVM models that the individual credit risk set up is assessed are
The step 3 includes following sub-step：
(3,1) defect of local extremum is easily trapped into when to parameter optimization for traditional drosophila optimized algorithm, the present invention
Drosophila algorithm is improved, improved method is as follows：
To find out an optimal drosophila position of current flavor concentration in traditional drosophila algorithm FOA, then drosophila group
The maximum drosophila position of the concentration is flown to, drosophila flies to destination and error occurs unavoidably, will cause to find optimal flavor concentration
The speed of value slows down.In traditional drosophila algorithm FOA, toss looks for optimal solution, seek approximate optimal solution it is general
Rate is smaller.In view of the above-mentioned problems, the present invention proposes a kind of improved drosophila algorithm (Improving fruit fly
Optimization algorithm, IFOA), effectively ensure the optimizing essence of algorithm while algorithm Searching efficiency can be improved
Degree.IFOA is to find out preferably n elite drosophila position of current flavor concentration, is weighted processing to these positions, obtains one
Individual optimal weighting position, allows drosophila group to be flown to the position.Let fly away operation carry out n time, select flavor concentration average it is optimal one
It is secondary to let fly away.Improved drosophila algorithm not only increases the probability for finding optimal solution, is also added the speed for finding optimal solution
It hurry up.
The step (3,1) specifically includes following sub-step：
Step A1, random initializtion drosophila group position X1With Y1, drosophila colony number is m, and drosophila population iterations is
N。
Step A2, due to the position where food can not be learnt in advance, so first estimation drosophila and the origin of coordinates apart from d
(i,:), then calculate flavor concentration decision content s (i,:), this value is the inverse of distance.
Step A3, searches out n elite drosophila of flavor concentration highest in this drosophila colony, place is weighted to its position
Reason, obtains the position of Weighted optimal
Step A4, allows drosophila colony to be flown to the Weighted optimal position by visionThis operation is performed n times, selects target letter
Average is optimal once lets fly away for number.
Step A5, retains the optimal flavor concentration value F of average and optimal weighting drosophila position.
Step 4 includes following sub-step：
(4,1) size and iterations of drosophila population in improved drosophila algorithm are initialized, the correlation of SVM models is selected
Parameter (SVM is used as kernel function using RBF RBFs).K=3 in SVM herein parameter K-CV methods, this method.Set
Drosophila population scale is 20, and iterative algebra is 50.
(4,2) are set up SVM training patterns and tested, and calculate fitness function, obtain drosophila group in iteration each time
Optimal parameter is combined in scale, and is recorded.
Step 5 includes following sub-step：
The individual credit risk based on IFOA-SVM that creditor's data that part is used as test are brought into step 4 is assessed
Model, carries out the assessing credit risks of creditor.
Brief description of the drawings
Fig. 1 is improved drosophila algorithm flow chart；
Fig. 2 is to set up IFOA-SVM model flow figures；
Fig. 3 IFOA iterativecurve figures；
The searching position figure of Fig. 4 parameter C and g drosophila individual；
Fig. 5 IFOA-SVM personal credit files predict the outcome；
Fig. 6 gridding method SVM model measurement collection predicts the outcome；
Fig. 7 GA-SVM model measurement collection predicts the outcome；
Fig. 8 FOA-SVM model measurement collection predicts the outcome.
Beneficial effect
Compared with prior art, the beneficial effects of the present invention are：Drosophila algorithm after improvement can improve algorithm and seek
The low optimization accuracy of algorithm is effectively guaranteed while excellent efficiency.With SVM, GA-SVM and FOA-SVM of gridding method search parameter
Evaluated effect contrasted, the results show IFOA-SVM models can obtain higher accurate in assessing credit risks
Rate.Estimated performance is improved, is the clear and definite loan customer credit of bank and other financial mechanism with stronger practicality, reduction loan
Risk provides effective foundation.
Embodiment
Technical solution of the present invention is illustrated below in conjunction with accompanying drawing.
Step 1), credit scoring model is quantified according to the personal data of creditor, it is determined that based on IFOA-SVM
The object function of people's credit risk assessment model；
Step 2), it is used for carrying out creditor the SVM models of credit evaluation based on SVMs foundation, discriminant function isWherein K (xi, yj) it is kernel function, b is constant, aiFor Lagrange factor i=1,2...n；
Step 3), drosophila algorithm is improved；
Step 4), using improved drosophila algorithm optimization SVM parameters, with improved drosophila algorithm to penalty factor and core
Function g carries out global optimizing, obtains the optimal solution of two parameters；Optimal solution parameter is updated to the SVM models in step 2
Training is practised, the individual credit risk assessment models based on IFOA-SVM are set up；
Step 5), by the part in step 1 be used as test creditor's data be brought into step 4) in based on IFOA-
SVM individual credit risk assessment models, and enter with SVM, GA-SVM and FOA-SVM of gridding method search parameter Evaluated effect
Row contrast.
The step 1) include following sub-step：
(1,1) 1000 borrower's information and its corresponding credit rating information are chosen, wherein 500 groups are used as training number
According to 500 groups are used for test data；
(1,2) choose the state of bank account, continue month, credit history, credit amount degree, savings account account
Family/bond, current working condition, marital relations, personal occupancy, real estate, the age, instalment plan, in this family silver
The existing amount of credit of row, available guarantor, whether telephone number whether there is register, as this 15 standard diagrams of foreign personage
As input variable, what is finally exported is its credit rating：1 represents risky, and 0 represents devoid of risk；
(1,3) object function is set to the root-mean-square error after SVM is predicted to data, and causes mean square error minimum
Change.
The step 2) include following sub-step：
(2,1) for given linear separability data setIt is linear in d dimension spaces
Discriminant function g (x)=wx+b, then can use hyperplane:It is normal vector that wx+b=0, which carries out w in sample separation, formula, and b is inclined
Move；In the case of linear separability, using yi(w·xi+ b) -1 >=0, i=1,2...n represents；
(2,2), can be by slack variable ξ when linearly inseparableiIt is incorporated into constraints,It is middle to introduce penalty factor to solve the problem；
(2,3) introduce kernel function when being nonlinear problemWherein σ ＞ 0；
(2,4) the SVM models that the individual credit risk set up is assessed are
The step 3) include following sub-step：
(3,1) following sub-step is specifically included when to parameter optimization for traditional drosophila optimized algorithm：
Step A1, random initializtion drosophila group position X1With Y1, drosophila colony number is m, and drosophila population iterations is
N；
Step A2, due to the position where food can not be learnt in advance, so first estimation drosophila and the origin of coordinates apart from d
(i,:), then calculate flavor concentration decision content s (i,:), this value is the inverse of distance；
Step A3, searches out n elite drosophila of flavor concentration highest in this drosophila colony, place is weighted to its position
Reason, obtains the position of Weighted optimal
Step A4, allows drosophila colony to be flown to the Weighted optimal position by visionThis operation is performed n times, selects target letter
Average is optimal once lets fly away for number；
Step A5, retains the optimal flavor concentration value F of average and optimal weighting drosophila position.
4th, the method as described in claim 1, it is characterised in that step 4 includes following sub-step：
(4,1) size and iterations of drosophila population in improved drosophila algorithm are initialized, the correlation of SVM models is selected
Parameter, SVM is used as kernel function using RBF RBFs；Using SVM parameter K-CV methods, K=3；Drosophila population rule are set
Mould is 20, and iterative algebra is 50；
(4,2) are set up SVM training patterns and tested, and calculate fitness function, obtain drosophila group in iteration each time
Optimal parameter is combined in scale, and is recorded；Fig. 3 is the iterativecurve figure that IFOA carries out parameter optimization to SVM, from Fig. 3
It can be found that IFOA looks for food initial stage in drosophila, drosophila hunting zone is to maximize, and the global optimizing ability of drosophila is also most strong, the 1st
Secondary iteration has just reached that more excellent fitness value is 8.6%.As drosophila is looked for food the increase of iterations, gradually approach optimal
Fitness value.Look for food the middle and later periods, with the enhancing of local search ability, have the adjustment of slight fitness value, finally the 29th
It is 7.6% that final fitness value (minimum mean square error) is reached at secondary iteration.Now obtain optimal C and the g ginseng of SVM models
Number.Representation parameter C successive dynasties drosophila is all in X-axis coordinate system range [55:100] it is distributed in.The hunting zone of drosophila it is very big and
Optimal drosophila individual distribution is more concentrated.The successive dynasties drosophila for being similarly represented as parameter g is all in Y-axis coordinate system range [35:75] in
Distribution, it is smaller compared with parameter C hunting zone, but possess very strong local search ability.Drosophila drosophila in search procedure
The location map of individual search parameter is as shown in Figure 4 in group.By can be calculated, now C=98.9373, g=46.4587,
IFOA can effectively search for C the and g parameters of supporting vector machine model as can be seen here.
The step 5) include following sub-step：
Part be used as test creditor's data be brought into step 4) in the individual credit risk based on IFOA-SVM comment
Estimate model, carry out the assessing credit risks of creditor, in order to embody the superiority of IFOA-SVM category of model Evaluated effects, respectively
Contrasted with SVM, GA-SVM Evaluated effect of gridding method search parameter.Predict the outcome as shown in Figure 5.In order to embody IFOA-
The superiority of SVM category of model Evaluated effects, the respectively assessment with SVM, GA-SVM and FOA-SVM of gridding method search parameter is imitated
Fruit is contrasted.It is 0.0474 that gridding method, which searches optimized parameter C for 2.2974, g, is predicted the outcome as shown in Figure 6.It is well-known
Genetic algorithm optimization SVM parameter effects are more excellent, itself in the present invention and FOA/IFOA population invariable number, evolutionary generation and parameter optimization
The parameter settings such as scope are essentially identical, genetic algorithm seek optimized parameter C be 47.6161, g be 50.6952, GA-SVM models
Predict the outcome as shown in Figure 7.Drosophila algorithm seek optimized parameter C be 71.3187, g be 48.1275, FOA-SVM models prediction
As a result it is as shown in Figure 8.The accuracy rate that IFOA-SVM is assessed is above that erroneous judgement number can be reduced after other models, drosophila algorithm optimization,
Significantly improve the degree of accuracy of assessing credit risks.
Claims (6)
1. a kind of individual credit risk appraisal procedure based on IFOA-SVM, it is characterised in that including：
Step 1), credit scoring model is quantified according to the personal data of creditor, it is determined that based on IFOA-SVM people's letter
With the object function of risk evaluation model；
Step 2), it is used for carrying out creditor the SVM models of credit evaluation based on SVMs foundation, discriminant function isWherein K (xi,yj) it is kernel function, b is constant, aiFor Lagrange factor i=1,2...n；
Step 3), drosophila algorithm is improved；
Step 4), using improved drosophila algorithm optimization SVM parameters, with improved drosophila algorithm to penalty factor and kernel function g
Global optimizing is carried out, the optimal solution of two parameters is obtained；The SVM models that optimal solution parameter is updated in step 2 carry out study instruction
Practice, set up the individual credit risk assessment models based on IFOA-SVM；
Step 5), by the part in step 1 be used as test creditor's data be brought into step 4) in based on IFOA-SVM's
Individual credit risk assessment models, and carried out pair with SVM, GA-SVM and FOA-SVM of gridding method search parameter Evaluated effect
Than.
2. the method as described in claim 1, it is characterised in that the step 1) include following sub-step：
(1,1) 1000 borrower's information and its corresponding credit rating information are chosen, wherein 500 groups are used as training data, 500
Group is used for test data；
(1,2) choose the state of bank account, continue month, credit history, credit amount degree, savings account/debt
It is certificate, current working condition, marital relations, personal occupancy, real estate, the age, instalment plan, existing in this bank of family
Amount of credit, available guarantor, whether telephone number whether there is registers, be foreign personage this 15 standard diagrams as defeated
Enter variable, what is finally exported is its credit rating：1 represents risky, and 0 represents devoid of risk；
(1,3) object function is set to the root-mean-square error after SVM is predicted to data, and mean square error is minimized.
3. the method as described in claim 1, it is characterised in that the step 2) include following sub-step：
(2,1) for given linear separability data set { xi,yi, i=1 ..., N, y ∈ { -1 ,+1 },D dimension spaces
In linear discriminant function g (x)=wx+b, then can use hyperplane:It is normal direction that wx+b=0, which carries out w in sample separation, formula,
Amount, b is skew；In the case of linear separability, using yi(w·xi+ b) -1 >=0, i=1,2...n represents；
(2,2), can be by slack variable ξ when linearly inseparableiIt is incorporated into constraints,It is middle to introduce penalty factor to solve the problem；
(2,3) introduce kernel function when being nonlinear problemWherein σ ＞ 0；
(2,4) the SVM models that the individual credit risk set up is assessed are
4. the method as described in claim 1, it is characterised in that the step 3 includes following sub-step：
(3,1) following sub-step is specifically included when to parameter optimization for traditional drosophila optimized algorithm：
Step A1, random initializtion drosophila group positionX1WithY1, drosophila colony number ism, drosophila population iterations is N；
Step A2, due to the position where food can not be learnt in advance, so first estimation drosophila and the origin of coordinates apart from d
(i,:), then calculate flavor concentration decision content s (i,:), this value is the inverse of distance；
Step A3, searches out n elite drosophila of flavor concentration highest in this drosophila colony, processing is weighted to its position,
Obtain the position of Weighted optimal
Step A4, allows drosophila colony to be flown to the Weighted optimal position by visionThis operation is performed n times, selects object function equal
Value is optimal once to let fly away；
Step A5, retains the optimal flavor concentration value F of average and optimal weighting drosophila position.
5. the method as described in claim 1, it is characterised in that step 4 includes following sub-step：
(4,1) size and iterations of drosophila population in improved drosophila algorithm, the related ginseng of selection SVM models are initialized
Number, SVM is used as kernel function using RBF RBFs；Using SVM parameter K-CV methods, K=3；Drosophila population scale is set
For 20, iterative algebra is 50；
(4,2) are set up SVM training patterns and tested, and calculate fitness function, obtain drosophila group scale in iteration each time
Middle optimal parameter combination, and record.
6. the method as described in claim 1, it is characterised in that step 5) include following sub-step：
Part is used as creditor's data of test and is brought into step 4) in individual credit risk based on IFOA-SVM assess mould
Type, carries out the assessing credit risks of creditor, in order to embody the superiority of IFOA-SVM category of model Evaluated effects, respectively with net
SVM, GA-SVM and FOA-SVM of lattice method search parameter Evaluated effect are contrasted.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710463836.9A CN107330781A (en) | 2017-06-19 | 2017-06-19 | A kind of individual credit risk appraisal procedure based on IFOA SVM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710463836.9A CN107330781A (en) | 2017-06-19 | 2017-06-19 | A kind of individual credit risk appraisal procedure based on IFOA SVM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107330781A true CN107330781A (en) | 2017-11-07 |
Family
ID=60195886
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710463836.9A Pending CN107330781A (en) | 2017-06-19 | 2017-06-19 | A kind of individual credit risk appraisal procedure based on IFOA SVM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107330781A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109062180A (en) * | 2018-07-25 | 2018-12-21 | 国网江苏省电力有限公司检修分公司 | A kind of oil-immersed electric reactor method for diagnosing faults based on IFOA optimization SVM model |
| CN109116833A (en) * | 2018-08-31 | 2019-01-01 | 重庆邮电大学 | Based on improvement drosophila-bat algorithm mechanical failure diagnostic method |
| CN111753083A (en) * | 2020-05-10 | 2020-10-09 | 北京工业大学 | Complaint report text classification method based on SVM parameter optimization |
| CN111832838A (en) * | 2020-07-24 | 2020-10-27 | 河北工业大学 | Method for predicting short-term wind power generation output power |
| CN112053223A (en) * | 2020-08-14 | 2020-12-08 | 百维金科(上海)信息科技有限公司 | Internet financial fraud behavior detection method based on GA-SVM algorithm |
| CN113379251A (en) * | 2021-06-11 | 2021-09-10 | 浙江工业大学 | IFOA-SVM-based high-voltage switch cabinet state evaluation method |
| TWI783387B (en) * | 2020-03-06 | 2022-11-11 | 日商日立系統股份有限公司 | Management support device, management support system, management support program, and management support method |
-
2017
- 2017-06-19 CN CN201710463836.9A patent/CN107330781A/en active Pending
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109062180A (en) * | 2018-07-25 | 2018-12-21 | 国网江苏省电力有限公司检修分公司 | A kind of oil-immersed electric reactor method for diagnosing faults based on IFOA optimization SVM model |
| CN109116833A (en) * | 2018-08-31 | 2019-01-01 | 重庆邮电大学 | Based on improvement drosophila-bat algorithm mechanical failure diagnostic method |
| CN109116833B (en) * | 2018-08-31 | 2021-04-16 | 重庆邮电大学 | Mechanical fault diagnosis method based on improved fruit fly-bat algorithm |
| TWI783387B (en) * | 2020-03-06 | 2022-11-11 | 日商日立系統股份有限公司 | Management support device, management support system, management support program, and management support method |
| CN111753083A (en) * | 2020-05-10 | 2020-10-09 | 北京工业大学 | Complaint report text classification method based on SVM parameter optimization |
| CN111832838A (en) * | 2020-07-24 | 2020-10-27 | 河北工业大学 | Method for predicting short-term wind power generation output power |
| CN111832838B (en) * | 2020-07-24 | 2022-03-01 | 河北工业大学 | Method for predicting short-term wind power generation output power |
| CN112053223A (en) * | 2020-08-14 | 2020-12-08 | 百维金科(上海)信息科技有限公司 | Internet financial fraud behavior detection method based on GA-SVM algorithm |
| CN113379251A (en) * | 2021-06-11 | 2021-09-10 | 浙江工业大学 | IFOA-SVM-based high-voltage switch cabinet state evaluation method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107330781A (en) | A kind of individual credit risk appraisal procedure based on IFOA SVM | |
| Wang et al. | DeepTrader: a deep reinforcement learning approach for risk-return balanced portfolio management with market conditions Embedding | |
| Chaudhuri et al. | Fuzzy support vector machine for bankruptcy prediction | |
| WO2019179403A1 (en) | Fraud transaction detection method based on sequence width depth learning | |
| CN107133865B (en) | Credit score obtaining and feature vector value output method and device | |
| CN111860658A (en) | Transformer fault diagnosis method based on cost sensitivity and integrated learning | |
| Alsubaie et al. | Cost-sensitive prediction of stock price direction: Selection of technical indicators | |
| CN102609714B (en) | Novel classification device and sorting technique based on information gain and Online SVM | |
| CN109670927A (en) | The method of adjustment and its device of credit line, equipment, storage medium | |
| Ng et al. | LG-Trader: Stock trading decision support based on feature selection by weighted localized generalization error model | |
| Chen | Using a hybrid evolution approach to forecast financial failures for Taiwan-listed companies | |
| Singh et al. | Cost-sensitive metaheuristic technique for credit card fraud detection | |
| Guo et al. | GRTR: Gradient rebalanced traffic sign recognition for autonomous vehicles | |
| Zhang | Analyzing Financial Market Trends in Cryptocurrency and Stock Prices Using CNN-LSTM Models | |
| CN105894023A (en) | Support vector data description improved algorithm based on clusters | |
| Li et al. | Forex prediction based on SVR optimized by artificial fish swarm algorithm | |
| Zheng et al. | Imbalanced problem in initial coin offering fraud detection | |
| Hsu et al. | Evaluating machine learning varieties for NBA players' winning contribution | |
| Loukeris et al. | Portfolio selection in hybrid Radial Basis Functions | |
| Li et al. | Denoising Diffusion Probabilistic Models and Transfer Learning for citrus disease diagnosis | |
| Wang et al. | Hybrid neural network based on GA-BP for personal credit scoring | |
| Petrlik et al. | Multiobjective selection of input sensors for svr applied to road traffic prediction | |
| Cui et al. | esearch on Credit Card Fraud Classification Based on GA-SVM | |
| Rahman et al. | A comparative study of undersampling and oversampling methods for flood forecasting in Bangladesh using machine learning | |
| Gong et al. | Predictive Classification Model Based on Vespa Comprehensive Detection System |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171107 |
|
| WD01 | Invention patent application deemed withdrawn after publication |