### Paper Menu >>

### Journal Menu >>

American Journal of Industrial and Business Management, 2013, 3, 429-434 http://dx.doi.org/10.4236/ajibm.2013.34049 Published Online August 2013 (http://www.scirp.org/journal/ajibm) 429 Study of Personal Credit Evaluation Method Based on PSO-RBF Neural Network Model* Shuai Li, Yuanmei Zhu, Chao Xu, Zongfang Zhou# School of Management and Economics, University of Electronic Science and Technology of China, Chengdu, China. Email: #zhouzf@uestc.edu.cn Received April 1st, 2013; revised May 10th, 2013; accepted June 10th, 2013 Copyright © 2013 Shuai Li et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. ABSTRACT Personal credit evaluation is the basic method for the commercial banks to avoid the consumer credit risk. On one hand, the credit behavior of individuals is complex; on the other hand the personal credit assessment system in our country is not sound, assessment methods are mostly objective, therefore, more and better scientific methods for credit risk as- sessment need to be introduced. This paper proposed a method for personal credit evaluation based on PSO-RBF neural network, which used PSO algorith m to optimize the parameters of RBF neural network, then applied the optimized RBF neural network in the personal credit evaluation. This method combined the global searching ability of PSO algorithm and the high effectiveness of local optimize o f RBF together, overcame the unstabitily of PSO algorithm and the draw- back of RBF which easily leads to local minimum. The result shows that the personal credit assessment method based on PSO-RBF neural network is highly accurate in classification and prediction, and is suitable in personal credit as- sessment and prediction. Keywords: Personal Credit Evaluation; PSO Algorithm; RBF Network; PSO-RBF Model 1. Introduction With the rapid development of Chinese economy, the personal credit is being taken more and more attention, which strongly ch allenges the credit evaluation system in our country. At present, the personal credit problem in our country has seriously affected the development of Chinese economy, therefore, to build scientific personal credit evaluation system, and to introduce in scientific personal credit evaluation is in urgent need. Regards to the commercial bank, personal credit assessment is to make comprehensive judgment and evaluate the possibil- ity of the borrower to return the money on time, accord- ing to the observation of the indexes which reflect the personal credit ability, and so as to make a decision on whether lend money [1]. In western countries, the theories and methods on credit risk evaluation are maturing, and many interdisci- plinary theories and methods are applied to credit risk evaluation [2]. Personal credit risk evaluation methods are separated into two types, statistic and non-statistic method, and the statistic method includes decision theory method, logic regression, linear and non-linear regression, K-nearest-neighbor estimation etc, non-statistics method includes mathematical programming, neural network, genic algorithm, expert system and classification tree method etc. [3]. Particle swarm optimization (PSO) [4] is a recently developed numerical method for optimization, which is simple, easy to apply and has a strong smart background, and it has been used in many fields such as function optimization, and pattern recognition. In 2007, Fu Peizhong, Ying Yixin [5] pioneered in apply PSO algorithm in determining the parameter of RBF neural network (connection weight, centers and widths of the hidden units), and checked its validity using simulation. RBF neural network is a three-layered BP network, which can approximate the continuous function with ar- bitrary precision. It is characterized by single best ap- proximation, no lo cal minimum, less calculation and fast learning, and is widely applied to pattern classification, system identification and functional approximation. At present, PSO-RBF neural network model [5] is applied to many industries, such as water quality assessment [6], short time traffic flow prediction [7], and boiler super- *Funding Project: Natural Science Foundation of China (71271043); The Special Research Foundation of Ph.D. Program of China (2011 0185110021) and Sichuan Province Science and Technology Support Project (2012SZ0001). #Corresponding autho r . Copyright © 2013 SciRes. AJIBM Study of Personal Credit Evaluation Method Based on PSO-RBF Neural Network Model 430 heat steam temperature control systems identification [8] etc. This paper built personal credit assessment model by using RBF neural network, and chose PSO algorithm to train and optimize the network, then built personal credit assessment model based on PSO-RBF neural network, and made a further comparative analysis of it with sing le RBF neural network, the second chapter of this paper provided the modeling solution; the third chapter proc- essed the sample data; the fourth chapter analyzed the result of simulation. 2. Basic Theory and Modeling Solution 2.1. RBF Neural Network Model RBF neural network [9] is firstly proposed by J. Moody and C. Darken, in the late 1980s, it is a three-layered BP network with a single-hidd en Layer. The first layer is the input layer, constructed with signal source; the second layer is the hidden layer, its action function is Gaussian function, the hidden layer proceed space mapping and transformation to the input information; the third layer is output layer, its action function is linear function, the output layer linear weighted the information from the neuron of the hidden layer, then output the result of the entire network. Its structure is as shown in Figure 1. The figure abov e is the FBF neur al network, specially, in this paper, 17 indicators are selected by the inputs layer in the article. Gaussian basis functions are used in Hidden layer. The hidden layer output results are weighted to get the final output, which means that the results of the assessment of the personal credit. As is known, generally, the RBF Neural Network has n inputs, m hidden node, 1 output; the action function of the radial base layer (hidden layer) is Gaussian function, the output of this layer neuron j is: 2 2 exp, 1,2,..., 2 i i i xc hi m h (1) Among them, x is an n sphere network input vector, ci is a center vector with sphere of x, is the basic width. The output of RBF neural network is: , among 1 m ii i yw them, w i is the network weight vector between the ith radius basic layer and output layer. When applying RBF neural network, three parameters need to be determined: the center vector of the Gaussian function ci, the basic vector i and the network weight wi. 2.2. Weighted PSO Algorithm Particle Swarm Optimization (PSO) [10-12], proposed by Kennedy and Eberhart in 1995, is a Cluster optimization calculation method, which can be used as neural network 1 x 2 x n x .... 2 h m h y 1 w m w 2 w 1 h Figure 1. The structure of RBF neural network based on gaussian function. training method. Its basic idea is to update the local op- timal 12 ,,, best d pppp and the global optimal 12 ,,, best d g gg g of the particles with the fitness (the fitness function normally is determined by the opti- mized function in practical issues) of each particle. The particle updates their speed and location, by using the following formula to dynamic trace the optimal of individuals and the global optimal. 112 2 1 ij ijj ijj ij vt wv tcrptx tcr g tx t (2) 1 ijij ij xtxt vt 1 (3) Among them, j = 1, 2, …, d; t is the number of itera- tion; r1, r2, are the ra ndom number uniform distributed in (0,1); c1, c2 are the accelerated factor, generally c1 = c2 = 2; w indicates the influence of the previous speed to the present. For better searching results, this paper adopts the im- proved weighted PSO algorithm. The weigh w maintains the balance of global and local searching capacity, high w inclines to global search, otherwise inclines to local searching [13]. This paper chose the following mean of square error function as the fitness function of the PSO algorithm. 2 ,, 11 1nm d ji ji ij MSEy y n (4) Among them, n indicates the number of the input variables, m indicate node number of the hidden layer, , d j i y indicates the idea output value, , j i y indicates the actual value. 2.3. RBF Neural Network Optimization Process Based on PSO Algorithm As is mentioned above, there are three important pa- rameters in RBF neural network, namely the cen ter of the Gaussian function cj, the basic vector i and the net- work weight wi. However, Currently, it is difficult to obtain the best values of the parameters of the network structure in theory, and the methods of training RBF Copyright © 2013 SciRes. AJIBM Study of Personal Credit Evaluation Method Based on PSO-RBF Neural Network Model Copyright © 2013 SciRes. AJIBM 431 neural network parameters have a major impact on the performance of the network, so we must select the ap- propriate parameters to improve the prediction perform- ance of the RBF neural network. This paper chooses par- ticle swarm optimization (PSO) to determine the pa- rameters of RBF neural network, so that they are no longer random. The optimized RBF neural network, compared to the independent PSO algorithm and RBF neural network algorithm, has a better evaluation effect on the assessment of the personal credit with higher pre- diction accuracy. Specific training steps [14] are as fol- lows: 1) To set the related parameters of the PSO algorithm, initialize the particles, including the v alue assignment for hidden center vector, base width vector, weight of the network; 2) Calculate the fitness value of each particle according to Formula (4), and make the current position of the particle as the individual maximum pbest, find out the particle of the minimum fitness value, and make it the initial gbest; 3) Compare fitness valu e of the current parti- cle with pbest, if the fitness value of the current one is sma ller, then upd ate pbest with the current fitness val ue; 4) For each particle, compare pbest with gbest, if pbest is better, then update gbest; 5) Update the speed and location of the particle according to formulas (2) and (3); 6) Repeat steps 4)-6), until the terminal condition is met, which is either meet the maximum iterations or the error accuracy requirement; 7) Set gbest as the parameter of the RBF neu- ral network. Specific processes are shown in Figure 2. 3. Sample Data and Preprocess 3.1. Sample Data and the Selection of Variables Because of the difficulty in collecting domestic credit data, this paper uses the personal credit data of a Ger- many bank, a total of 1000 data. Considering the specific circumstance of Germany, removing too much index [15] will influence the effect of the model, therefore this pa- per removed two indexes, “the purpose of loan” and “na- tionality”, which have little influence to the model. Spe- cific indexes and quantification method are shown in Table 1 [16]. First, this paper using stratified sampling, divided the 1000 sample clients into “good credit” and “bad credit”, with 700 and 300 respectively, and for each sample con- tains 17 attribute indexes. Then, 350 “good credit” and Original data and the preprocessing Building the RBF neural network model Optimizing the parameters using PSO Initial the speed, the location of the particles Calculating the fitness value of the particles Updating the speed and location of the particles Updating P b est with the current value The current value is better N The current value is better Updating g b est with the current value Make the optimal particles as the p arameter of the RBF Complete the model training Applying the PSO-RBF model in testing the rest samples Meet the terminal condition N Y N Y Out pu t Y Figure 2. The flow chart of PSO-RBF neural model. Study of Personal Credit Evaluation Method Based on PSO-RBF Neural Network Model 432 Table 1. Variable and its quantification. Indexes Variable Definition Age X1 Actual value Marriage X2 1 = single; 2 = mar ried Suppor t i n g f a mil y membe rs X3 Actual value Occupation X4 1 = unemployed/manual workers, non-resident; 2 = non-proficient worker, resident; 3 = proficient worker/officer; 4 = manager/i ndepe nden t en tr epr e neur s Year of working X5 1 = unemployed; 2 = less than 1 year; 3 = 1 - 4 years; 4 = 4 - 7 years; 5 = more than 7 year Housing condition X6 1 = rent; 2 = owned; 3 =free housing Year of live in current house X7 Actual value Installment to deposable disposable income rate X8 Actual value assets X9 1 = real estate; 2 = if n o t 1 : a g reement of public construction savings/life insurance; 3 = if not 1 or 2: Automobile or other; 4 = vain Current payment account status X10 1 = less than 0 mark; 2 = 0 - 200 dollar; 3 = more than 200 dollar or Salary contract has been signed for at least a year; 4 = no payment account The rest plan for the installment X11 1 = bank; 2 = stock ; 3 = no Debt amount X12 Actual value Saving account / bo nds X13 1 = less than 100 mark; 2 = 100 - 200 dollars; 3 = 500 - 1000 dollars; 4 = more than 1000 dollars; 5 = no saving account/bonds Loan Period X14 Actual value Credit record X15 0 = no bad credit record; 1 = has overdue payment record/other bad cr e di t record; 2 = overdue payment; 3 = has late payment record; 4 = no credit record/credit record is no in this bank Existing loan project number in this bank X16 Actual value other note debtor/guarantor X17 1 = no; 2 = joint applicants; 3 = secured 150 “bad credit” from the sample are selected as the training sample; the rest 500 samples are used in testing the model, to exam the classification effect of the model. 3.2. Normalizing the Data To accelerate the convergence of the model and reduce the effect to the network classification caused by the un- balance of the data, first normalize the training sample and the test sample. Divide the adopted 17 explanatory variables into discr e te va riable and continuou s variable. For the discrete variable, use the Minimum and maxi- mum normalization method to process. min max min xx X x x (5) Among them, 0,1X is the value of the variable after normalizing, xmin, xmax are the maximum and mini- mum value of the variables. For the continuous variables, through observing the distribution of the variables , we found that the amount of the loan x12,and the time of the loan x14 both approxi- mately obey the law of normal distribution, that is 2 ~,xNu , and convert them into [0,1]. ' x u X (6) Among them, 2 2 1 2 t x x edt is the standard normal distribution, through converting, we can get the corresponding probability. 4. Simulation and the Result Analysis In the process of designing RBF neural network, the key is the determining of the number of the hidden neurons, because there are many parameters in this paper, when we set the number of the hidden neurons as 3, the result meet the setting condition, and the input parameters has 17 variables, then the structure or RBF network is 17-3-1. To obtain better training result, this paper chose the im- provement PSO algorithm with the inertia weight, the inertia weight w = 0.1, the learning factor c1 c2 are both 2, and, the maximum iterations tmax is 1500, which is the termination condition of the algorithm. To keep the con- Copyright © 2013 SciRes. AJIBM Study of Personal Credit Evaluation Method Based on PSO-RBF Neural Network Model 433 sistency of the data, we chose the 500 sample data in the testing sample as the fundament of the assessment, when evaluating the credit of the creditor, use 0.5 as the critical value, the one whose predicted value is higher than 0.5 is viewed as “good credit” client, otherwise is “bad credit” client, the result of the comparison is shown in Table 2. In practice, there ar e two types of errors: type one and type two, th e former is to misjudge the good credit clien t as bad credit client and rejects their loan application; the latter is to misjudge the bad credit client as good credit client and accept their loan application. Therefore type two errors cause more damage to the bank, and the false rate of it should be reduced. From the table above, it can be seen that PSO-RBF neural network when compared with the pure RBF neural network model, both the type one error, and type two error rate are reduced , therefore the total false rate is reduced, what’s more, the type two error rate is reduced from 8.67% to 4%, namely, the probability of misjudging “bad credit” client as “good credit” client is reduced , which has more practical value for the commercial bank to avoid the personal credit risk. To test the perdition accuracy of the model, after 1500 iterations, we get the following evaluation error char t As the below Figure 3 shown, evaluation error rate of PSO-RBF model is apparently lower than RBF model, which means PSO-RBF model is effective and of better accuracy. If adopting the RBF model directly, because the initial value of the network is random, the evaluation effect is poor. In conclusion, first adopt PSO algorithm to adjust and optimize the parameter of the RBF model, and reduce the inner error of the model, and then apply the Table 2. Comparison between RBF neural networ k and PSO-RBF neural network model prediction accuracy. Model Actual value Predicted value 1.00 0.00 Sample number Accuracy (%) False rate (%) 1.00 317 33 350 90.57 9.43 0.00 13 137 150 91.33 8.67 RBF Total 330 170 500 90.80 9.20 1.00 326 24 350 93.14 6.86 0.00 6 144 150 96.00 4.00 PSO-RBF Total 332 168 500 94.00 6.00 Figure 3. Evaluation error curve of PSO-RBF model and RBF model. Copyright © 2013 SciRes. AJIBM Study of Personal Credit Evaluation Method Based on PSO-RBF Neural Network Model 434 optimized model in the personal credit evaluation, the evaluation result is more accurate than the single RBF model. 5. Conclusions This paper adopting PSO algorithm in the training of RBF neural network, proposed PSO-RBF neural network method, used PSO algorithm to proceed global dynamic searching, and used RBF neural network to proceed local optimizing, finally got the PSO-RBF neural network. This paper first adopted PSO-RBF neural network in personal credit risk evaluation, and through simulation and the empirical study of the 1000 personal credit sam- ple data from a Germany commercial bank, indicated that this method when compared with the pure RBF model, showed higher classification and evaluation accuracy, re- duced the type two error, and effectively reduced the personal credit risk that the commercial bank facing. Therefore PSO-RBF neural network is applicable in per- sonal credit risk evaluation. There is potential im- provement in this paper, mainly in three aspects: 1) According to reducing the type two error, how to choose the fitness function of the particles in the PSO algorithm; 2) The number of the hidden neurons in the RBF model can be found through other optimizing algorithm, so as to improve the performance of the model; 3) Because there are too many input variables, the convergence of the model is comparatively slow, the time performance is poor. REFERENCES [1] J. Z. Dai, “Exploration on the building of Personal Credit Risk System,” Financial Practice, Vol. 15, No. 3, 2002, p. 11. [2] Z. F. Zhou, et al., “The Study of the Evolution Mecha- nism and Evaluation Methods of Emerging Technology Corporate Credit Risk,” Science Press, Beijing, 2010. [3] W. E. Henley, “Statistical Aspects of Credit Scoring,” The Open University, Milton Keynes, 1995. [4] J. Kennedy and R. C. Eberhart, “Particle Swarm Optimi- zation,” Proceedings of International Conference on Neural Networks, Orlando, 1997, pp. 1942-1948. [5] P. Z. Fu and Y. X. Ying, “The Study of Neural Network Based on PSO-RBF and Its Application in Credit,” Chi- nese Journal of Scientific Instrument, Vol. 8, No. 1 2007, pp. 464-466. [6] D. S. Wang, S. H. Li and X. P. Zhou, “The Method and Application of Raw Water Quality Evaluation Based on PSO-RBF Neural Network Model,” Journal of Southeast University (Natural Science Edition), Vol. 41, No. 5, 2011, pp. 1019-1023. [7] M. F. Feng and J. C. Lu, “Use Particle Swarm to Opti- mize RBF Neural Network in Short-Term Traffic Flow Prediction,” Computer Simulation, Vol. 27, No. 12, 2010, pp. 323-326. [8] B. X. Xiao, X. W. Wang and Y. F. Liu, “Simulation Re- search of Boiler Superheat System Identification Using MPSO-RBF Optimization Strategy,” Journal Of System Simulation, Vol. 19, No. 6, 2007, pp. 1382-1386. [9] L. L. Li, “RBF Neural Network Integration Research and the Application in the personal Credit Assessment,” Shandong Normal University, Jinan, 2008. [10] S. Yin and M. H. Jiang, “Personal Credit Assessment Combination Forecast Model Based on PSO,” Economic Research Guide, Vol. 3, No. 14, 2208, pp. 83-86. [11] M. H. Jiang and X. C. Yuan, “The Construction and Ap- plication of PSO-SVM Model in Personal Credit Assess- ment,” Journal of Management, Vol. 5, No. 4, 2008, pp. 511-516. [12] L. Y. Yu, “Evaluation Prediction Model Research on Commercial Bank Credit Risk,” Journal of Management, Vol. 6, No. 5, 2003, pp. 46-52. [13] M. H. Jiang, “Research on Combination Forecast Method for Commercial Bank Personal Credit,” Harbin Institute of Technology, Harbin, 2006. [14] H. Y. Tang and Z. G. Xu, “Neural Network Based on PSO-RBF and Its Application. Metallurgical Automa- tion,” 2006, pp. 464-466. [15] J. Wang, “The Establishment of Personal Credit Evalua- tion Index in Consumer Credit,” Rural Finance Ressarch, Vol. 21, No. 1, 2001, pp. 48-50. [16] H. Z. Huang, Z. F. Zhou and J. K. Yu, “ILMBP Neural Network Model and Its Application in the Personal Credit Assessment,” Managerialist, Vol. 3, No. 10, 2010, pp. 1-3. Copyright © 2013 SciRes. AJIBM |