Microsoft Word - cet-01.docx


 

 CHEMICAL ENGINEERING TRANSACTIONS  
 

VOL. 46, 2015 

A publication of 

 

The Italian Association 
of Chemical Engineering  
Online at www.aidic.it/cet 

Guest Editors: Peiyu Ren, Yancang Li, Huiping Song  
Copyright © 2015, AIDIC Servizi S.r.l.,  
ISBN 978-88-95608-37-2; ISSN 2283-9216                                                                                

A New Risk Forecasting Model of Financial Information 
System Based on Fitness Genetic Optimization Algorithm and 

Grey Model 
Yue Zhao 

School of Computer Science and Engineering, Jilin Jianzhu University, Changchun, Jilin, China, 130118 . 
zhaoyue3762@163.com 

The problems of security and risk are the main obstacles for the development of electronic bank. Along with the 
gradual advance of modernization about bank business, based on the existing financial risk, the risk of 
information system is added and close to the business of electronic bank. The business of electronic bank 
differing with that of tradition bank is a combination of information technology and electronic commerce. 
Therefore, the risk of its business is related to the level of information technology, information systems and the 
conditions of hardware equipment. In this paper, genetic algorithm is introduced to optimize corresponding 
parameters of grey forecasting model and then the improved grey model can be used in early warning of the risk 
in the information system of bank. Qualitative and quantitative forecasting methods are combined with grey 
model to obtain predicted value of risk index of the bank’s information system. Finally, the result of experiment 
states that the proposed method is reliable and exact to promote the development of business in electronic bank 
in the future. 

1. Introduction 

Electronic bank always suffers with the problems of security and risk. The business of electronic bank differing 
with that of tradition bank is a combination of information technology and electronic commerce. Therefore, the 
risk of its business is related to the level of information technology, information  systems and the conditions of 
hardware equipment. If the supervision mechanism of electronic bank is the same as the traditional one, it will 
become difficult for the supervision of risk in electronic bank. Therefore, it is also difficult for Banking Regulatory 
Commission to assess and supervise the risk of electronic bank. 
There are many existing studies on assessment of risk in electronic bank. Long et al. (2008) first divided the 
risks of electronic bank into internal risk and external risk and used fuzzy comprehensive evaluation method to 
assess the internal risk. For three main risks including the risks of laws, market and operation, Li (2001) 
evaluated these three risks and proposed corresponding suggestion. Fuzzy logic method is applied by Maher et 
al. (2008) to assess the security of electronic bank, which is based on fuzzy logic system and IF-THEN 
operation. According to the framework of COBIT, Guan (2009) developed a classified management method of 
electronic bank to evaluate its risk. This method achieved a satisfactory result. Fuzzy comprehensive evaluation 
method as a famous tool is also used by W ang (2009) to evaluate the risk of electronic bank and propose some 
corresponding measures to prevent its risk. Welch (1999) defined risk management and  researched it of 
electronic bank. Lian (2007) stated that data stream has five levels including application layer, middleware, 
database management, operation system and network which may affect information security. Then, hierarchy 
method should be introduced to assess the model. Sulivan (2000) developed the impact of electronic bank on 
traditional bank and analyzed this impact from the perspectives of acceptability and risk. The risk of coffers in 
electronic bank is proposed by Spivey (2001).  
In practice, the application of data in electronic bank is always not exact. Therefore, the early warning of the risk 
in electronic bank should be exact to guarantee corresponding benefits. Grey model has been widely applied in 
this field (Jiang, 2004; Lee 1986; Lin and Yang, 2003; Tien, 1996; Liu et al, 2001, Jie et al, 2004). Deng (2009) 
first proposed grey system theory to deal with incomplete data. Then, this method is developed by Liu et al. 

                               
 
 

 

 
   

                                                  
DOI: 10.3303/CET1546045

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Please cite this article as: Zhao Y., 2015, A new risk forecasting model of financial information system based on fitness genetic optimization 
algorithm and grey model, Chemical Engineering Transactions, 46, 265-270  DOI:10.3303/CET1546045  

265



 

(2010) to form a way of addressing corresponding information. However, there are few existing studies about 
early warning of the risk in electronic bank. 
Therefore, in this paper, genetic algorithm is introduced to optimize corresponding parameters of grey 
forecasting model and then the improved grey model can be used in early warning of the risk in the information 
system of bank. Qualitative and quantitative forecasting methods including expert scoring and mathematic 
modeling are introduced. First, based on the real case of the business in electronic bank, an attribute system is 
constructed. Then, different experts provide corresponding score on each attribute according to their 
preference, which will be normalized and then considered as samples. Moreover, grey model with genetic 
algorithm is proposed to forecast the risk of information system in electronic bank to obtain corresponding risk 
value. Finally, the result of experiment states that the proposed method is reliable and exact to promote the 
development of business in electronic bank in the future. 

2. Grey model 

Grey system theory is proposed by Deng to firstly develop a method used to manipulate data where partial 
information is known. Then, we will review some basic concepts of positive accumulated generation operator 
and regressive generation operator. 

The initial series is given as (0) (0) (0) (0){ (1), (2), ( )}X x x x n , where n is dimension, X(0) is grey series 

and (0)1, ( ) 0 n x i . 

Then we redesign new series (1) (1) (1) (1){ (1), (2), ( )}X x x x n , where  

(1) (0)

1

( ) ( )


 
k

i

x k x i  

(1) 

X(1) is one positive accumulated generation series of X(0). That is, (1) (0)X AGOX . If r positive 

accumulation is applied based on equation (1), it can be obtained that ( ) ( 1)

1

( ) ( )




 
k

r r

i

x k x i . 

If (0) (1) (1)( ) ( 1)   x x k x k , X(1) is one positive regressive generation series of X(0). That is, 
(1) (0)
X IAGOX . If r positive regression is applied based on equation (1), it can be obtained that 

( 1) ( ) ( )
( ) ( 1)


   

r r r
x x k x k . 

3. Genetic algorithm 

In general, genetic algorithm can be used to assess corresponding parameters a and b, the detailed process is 
demonstrated as follows: 
step 1, binary coding. We first determine the threshold of developing coefficient a and grey action criterion b. 
Then, genetic algorithm is used to code a and b to acquire the initial population which is the demand of genetic 
algorithm. 
step 2, construction of fitness function. The terminal condition of finding optimization parameter depends on the 
value of fitness function judged by us. If this value accords with the demand of us, prediction accuracy of grey 
model reaches the request. Meanwhile, this genetic algorithm can be terminated. Thus, parameters a and b are 
the optimization values after the process of decoding. Then these optimization parameters can be applied to 
the process of modeling the grey forecasting model. However, if the fitness value does not reach the request, 
we can go to the step 4. 
Value of fitness degree of genetic algorithm can be measured by the following equation. 

21
ˆ( )  i iMSE x x

N
 

(2) 

xi Is the real value, ˆ
i

x is the prediction value, and N is the total number of samples. The smaller vale the result 
is, the more exact the genetic algorithm is. 
When fitness is lower than average fitness, performance of this individual is not good. Thus, the major 
crossover rate and mutational rate are adopted, and vice versa. That is, when the value of fitness is closer to 
the biggest one, the crossover rate and mutational rate are smaller. Especially, when the value of fitness 
equals to the biggest one, the crossover rate and mutational rate equals to zero. This method is more suitable 

266



 

for the later evolution of population. Ai that time, the performance of each individual in population is good and 
thus it is not suitable for changing individual to break this good performance.  
Therefore, in this paper, genetic algorithm is improved adaptively and he crossover rate and mutational rate are 
demonstrated in the following. 

, ,

max ,

( )


  
 

 


’ ’

’

ci cj

ci avg avg

avgc

avgi

p p
p f f f f

f fp

f fp

 

(3) 

max
,max

( )
  
 




’

mi mj
avg

mi

avgm

avg
i

p p
f fp f f

f fp
f f

p

 

(4) 

step 3, calculation of fitness function. Based on the application of proposed algorithm in the risk assessment 
system of electronic bank and the preference of risk, we reset the parameters as follows. 
Suppose the size of population (M = 200), the number of terminal evolution (T = 50), the bigger crossover rate 
(Pc=0.975), the smaller mutational rate (Pm=0.001).  
step 4, selection, crossover and mutation. Through the selection, crossover and mutation of genetic algorithm, 
the population can be obtained. Then we can go to step 2 to judge the value of fitness. 
As the mentioned algorithm, the estimated value of the important parameters a and b can be acquired. 

4. Numerical example and result analysis  

Firstly, based on the existing reference, 5 experts which are from different fields inclu ding computer, finance 
and management are invited to give 9 attributes. Then, they provide the score of the risk about electronic bank 
on each attribute demonstrated in Table 1. Because of the limitation of content, the data of 2012 and 2013 is 
omitted.  
Then according to the comments in Table1, we get the normalized three -level index evaluation m atrix 
L = (Lij). Multiply three-level attribute evaluation m atrix and three-level attribute comprehensive weights 

to have three-level attribute evaluation results. Based on  
j j

i i
E T V , the V = {90,80,70,60,50}, we 

can calculate the risk assessment indexes of the 9 two-level index layers from  2011 to 2012. The i 
means indicators content and i = 1,2,…,9. 
We get a and b which are the background parameters of optimized gray forecasting model. According to the 
degree of their impact on results, MATLAB is used to obtain parameters 0.031124 and 13411.2135. 
Based on the above risk attributes, the gray forecasting model will be used to forecast the risk asses sment 
indexes of the next four periods: the first half and the second half of 2014 and 2015 showed in Table 3.  
Through comparison the result with the practical situation, to a large extent electronic bank realizes 
virtualization. But it leads to major increase of risk about business of electronic bank, management risk, 
especially, the selection risk of internet technology and artificial risk. It is further demonstrated that this work 
depends on artificial operation and need maintenance and management of ban kers. Therefore, the essential 
supervision and inspection of business and management can decrease these risks. This system of supervision 
and inspection can be divided into three levels. First, the main work is to inspect whether some 
countermeasures are used to comply with the corresponding laws and regulations. Meanwhile, essential 
disclosure must be executed to advertise non deposit investment instruments. Second, agreement between 
bank and some relevant person such as clients and technology vendors must  be inspected that whether it 
provides corresponding power and obligation. Then, it must be inspected that whether the laws of digital 
signature and authentication institutions are considered and an agreement between bank and a third party is 
included in corresponding documents. Thirdly, it must be inspected that whether corresponding confidentiality 
agreement is set to determine burden sharing. Because risks of control, operation responsibility and conflict 
overseas are less impact on the bank system, its risk attribute becomes level off. 
 
 

267



 

Table 1: Technical risk evaluation results of electronic banks  

1st grade 2nd grade 3rd grade Comment 
The first half 

 of 2011 
The second half 

 of 2011 

Score Normalization Score Normalization 

The risk 
of 

electronic 
bank 

Organization 
risk 

Planning 
and 

organization 

Excellent 3 0.6 1 0.2 
Good  2 0.4 4 0.8 
General  0 0 0 0 
Qualified  0 0 0 0 
Unqualified 0 0 0 0 

The 
development 
strategy of 
business 

Excellent 0 0 1 0.2 
Good  3 0.6 4 0.8 
General  2 0.4 0 0 
Qualified  0 0 0 0 
Unqualified 0 0 0 0 

The risk 
management 
of electronic 

bank 

Excellent 1 0.2 1 0.2 
Good  2 0.4 4 0.8 
General  2 0.4 0 0 
Qualified  0 0 0 0 
Unqualified 0 0 0 0 

Management 
risk 

information 
security 

management 
of electronic 

bank 

Excellent 0 0 1 0.2 
Good  5 1 4 0.8 
General  0 0 0 0 
Qualified  0 0 0 0 
Unqualified 0 0 0 0 

Security 
strategy 

Excellent 1 0.2 1 0.2 
Good  4 0.8 4 0.8 
General  0 0 0 0 
Qualified  0 0 0 0 
Unqualified 0 0 0 0 

Human 
resource 

management 

Excellent 0 0 2 0.4 

Good  5 1 3 0.6 
General  0 0 0 0 
Qualified  0 0 0 0 

 

  Unqualified 0 0 0 0 

Control risk 

Internal 
control 

organization 
of electronic 

bank 

Excellent 2 0.4 1 0.2 
Good  3 0.6 4 0.8 
General  0 0 0 0 
Qualified  0 0 0 0 
Unqualified 0 0 0 0 

The control 
to service 
provider 

Excellent 1 0.2 1 0.2 
Good  4 0.8 4 0.8 
General  0 0 0 0 
Qualified  0 0 0 0 
Unqualified 0 0 0 0 

Audit of 
electronic 

bank 

Excellent 0 0 1 0.2 

Good  5 1 4 0.8 
General  0 0 0 0 
Qualified  0 0 0 0 
Unqualified 0 0 0 0 

 
 
 

268



 

Table 2: The risk assessment index  

 1j  2j  3j  4j  5j  6j  

1

j
E  81.703 82 84.851 84.561 85.703 89.280 

2

j
E  80.652 82.438 84.032 82.000 84.000 88.652 

3

j
E  82.331 82.000 84.000 83.851 84.830 88.000 

4

j
E  82.614 80.693 85.141 82.000 84.000 88.000 

5

j
E  82.661 82.911 84.854 87.145 84.855 89.517 

6

j
E  81.232 82.000 84.000 86.927 85.232 88.000 

7

j
E  82.236 82.407 84.636 86.565 84.000 88.734 

8

j
E  80.239 82.000 84.470 82.000 84.000 88.239 

9

j
E  80.000 83.382 81.379 82.000 84.000 88.000 

 

Table 3: The prediction results 

 The first half of 
2014 

The second half 
of 2014 

The first half of 
2015 

The second half of 
2015 

1

j
E  82.15 89.46 90.21 88.25 

2

j
E  82.27 88.46 87.26 86.42 

3

j
E  84.43 90.35 90.24 89.65 

4

j
E  87.42 85.21 85.33 89.21 

5

j
E  83.21 87.31 87.24 90.13 

6

j
E  85.18 85.57 88.17 89.34 

7

j
E  86.44 85.28 87.21 86.42 

8

j
E  87.62 84.61 88.13 87.59 

9

j
E  86.44 85.14 86.32 89.56 

 
 

269



 

Reference 

Aburrous M, Hossain M A, DahalKeshav T F, 2008, Intelligent quality performance assessment for E-banking 
security using fuzzy logic. International Conference on Information Technology: New Generations, 420 -425. 

Deng J L, 2009, The primary methods of grey system theory. Wuhan: Huazhong university of science and 
technology press. 

Guan L L, 2009, Research on risk assessment system of internet bank based on COBIT, Commercial Times, 
100-101. 

Jiang Y Q, 2004, Applying grey forecasting to predicting the operating energy performance of air cooled 
watercolors. International Journal of Refrigeration, 27(4): 385-392. 

Jie J X, Song B F, Liu D X, 2004, Gray correlation analysis method for scheme selection decision of aircraft 
top-layer design. Journal of Systems Engineering, 19(4): 350-354. 

Lee C, 1986, Grey system theory wins application on earthquake forecasting. Journal of Seismology, 4(1):  
27-31. 

Li D, 2001. The development and supervise of internet bank. Journal of Region Financial Research, 6:  4-7. 
Lian, 2007, Functionally and usability in design for e-banking Service. Internet with Computer, 7:12-13  
Lin C T, Yang S Y, 2003, Forecast of the output value of Taiwan’s optoelectronics industry using the grey 

forecasting model. Technological Forecasting and Social Change, 70(2): 177-186. 
Liu S F, Dang Y G, Fang Z G, Xie N M, 2010, Grey system theory and its application, Beijing: Science press. 
Liu S F, Zhao L, Wang Z Y, 2001, A new method of venturous capital pricing, Chinese Journal of Management 

Science, 9(2): 22-26. 
Long H Y, Xi Z F, Song G X, 2008, Interior risk of E-banking based on fuzzy colligate judge analysis. Modern 

Electronic Technique, 24: 151-157. 
Spivey J, 2001, Banks vault into online risk. Security Management 45 (1), 132-138. 
Sullivan R, 2000, How has the adoption of Internet banking affected performance and risk in banks. Financial 

Industry Perspectives, Federal Reserve Bank of Kansas City (December), 1-16. 
Tien T L, 1996, Study on prediction and decision making by grey theory, National  Cheng Kung University. 
Wang R H, 2009, Research on risk assessment of internet bank based on fuzzy colligate judge, Modern 

Electronic Technique, 11: 151-162. 
Welch B, 1999, Electronic Banking and Treasury Security. New York: Woodhead Punishing, 33-35. 

270