Download - Academy Publisher
Download - Academy Publisher
Download - Academy Publisher
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
ISBN 978-952-5726-09-1 (Print)<br />
Proceedings of the Second International Symposium on Networking and Network Security (ISNNS ’10)<br />
Jinggangshan, P. R. China, 2-4, April. 2010, pp. 073-076<br />
Improving Prediction Of Model Gm (1, 1) Based<br />
On Class Ratio Modeling Method<br />
Qin Wan 1 , Yong Wei 2 , and Shuang Yang 3<br />
1 College of Mathematics and Information,<br />
China West Normal University, NanChong, SiChuan, 637009, China<br />
2 College of Mathematics and Information,<br />
China West Normal University, NanChong, SiChuan, 637009, China<br />
3 College of Mathematics and Information,<br />
China West Normal University, NanChong, SiChuan, 637009, China<br />
wanqin1014@126.com, 3306866@163.com,yangshuang_cwnu@163.com<br />
Abstract—This article introduces a new method of how to<br />
find suitable class ratio according to the class ratio<br />
modeling method thought. It causes the class ratio modeling<br />
method to be used on the non-steady primitive sequence<br />
which has non-homogeneous grey index law directly. And it<br />
both extends the application scope of the class ratio<br />
modeling method, and avoids the tedious data pretreatment<br />
process effectively while improving prediction precision of<br />
model GM (1, 1).<br />
Index Terms — Class ratio modeling , GM (1,<br />
1) ,Weakening , Buffer operator<br />
I. INTRODUCTION<br />
Since Mr. Deng Julong has proposed the grey system<br />
theory, the application of grey model spreads many<br />
domains. Grey model has more advantages compared to<br />
traditional prediction method because that grey model<br />
has the characteristics of few sample data required, easy<br />
calculation, and high prediction accuracy in short terms<br />
etc. However in practical application, people discovered<br />
that model GM (1, 1) is suitable for slowly increasing<br />
data, but its fitting effect with quickly increasing data is<br />
unsatisfactory. Professor Liu Sifeng had theoretically<br />
proven that the applicable scope of development<br />
coefficient a is limited to ( − 2 , 2 ) , and the effective<br />
range of it is narrower [ 1]<br />
. Therefore, many scholars have<br />
made improvement and optimization on grey model from<br />
different angles. For instance, many references expand<br />
the applicable scope of development coefficient a by<br />
reconstructing background value or optimizing grey<br />
derivative to improve simulating and predicting precision<br />
of grey model, and achieve good results.<br />
Because the process of obtaining a by solving grey<br />
differential equation is just making whitenization<br />
estimation on a,the author in reference [2] puts forward<br />
class ratio modeling method of single consequence in<br />
[Key research projects] A project supported by scientific research fund<br />
of Sichuan Education Department.(2006A007) and basic application<br />
research fund of Sichuan (2008JY0112).,and a project supported by<br />
china west normal university item(08B032).<br />
Wan Qin: teaching assistant of College of Mathematics and<br />
Information , China West Normal University. major study is Grey<br />
System Analysis. Tel:18990874811. E-mail: wanqin1014@126.com<br />
grey system based on this idea.The primary data<br />
sequence is not suitable for establishing grey model<br />
directly if it could not pass feasibility test, so we can<br />
make data preprocessing such as carrying buffer<br />
[3]<br />
operator on the primitive behavior data sequence<br />
before establishing model GM(1,1) on it. Reference [4]<br />
propose using weakening buffer operator to affect the<br />
primitive behavior data sequence which possesses<br />
characteristics: the front part grows (weakens)<br />
excessively quickly and the latter part grows (weakens)<br />
excessively slowly; Reference [5] propose using<br />
strengthening buffer operator to affect to the primitive<br />
behavior data sequence which has subsequent<br />
characteristics: the front part grows (weakens)<br />
excessively slowly and the latter part grows (weakens)<br />
excessively quickly. Buffer operators can effectively<br />
eliminates the disturbance affect to the primary data<br />
effectively in the modeling and forecasting process, and<br />
improve the predicting precision of model GM (1, 1). For<br />
primary data sequence which has above characteristic,<br />
this paper introduces a new method of how to find<br />
suitable class ratio according to the class ratio modeling<br />
method thought. It causes the class ratio modeling<br />
method to be used on the non-steady primitive sequence<br />
which has non-homogeneous grey index law directly.<br />
And it both extends the application scope of the class<br />
ratio modeling method, and avoids the tedious data<br />
pretreatment process effectively while improving<br />
prediction precision of model GM (1, 1).<br />
II.<br />
SUMMARY OF THE CLASS RATIO MODELING<br />
METHOD<br />
[2]<br />
Definition As for monotone sequence<br />
X = { x(<br />
k ) x(<br />
k ) > 0, orx ( k ) < 0, k = 1,2,3.... n}<br />
,then<br />
x(<br />
k −1)<br />
σ ( k)<br />
= , k ∈ K = { 2,3,... n}<br />
is called the back<br />
x(<br />
k)<br />
class ratio (hereinafter using class ratio) at k point of<br />
X , and x(<br />
k )<br />
τ ( k ) = , k ∈ K = { 2,3,... n}<br />
is called<br />
x(<br />
k − 1)<br />
the front class ratio at k point of X .Obviously, the back<br />
class ratio and the front class ratio both are the reciprocal<br />
value of each other, and both are positive numbers.<br />
© 2010 ACADEMY PUBLISHER<br />
AP-PROC-CS-10CN006<br />
73