Title :
Fuzzy regression analysis by entropy
Author :
Kao, Chiang ; Lin, Pei-Huang
Author_Institution :
Dept. of Ind. & Inf. Manage., Nat. Cheng Kung Univ., Tainan, Taiwan
Abstract :
To construct a regression model for fuzzy numbers, this paper decomposes a fuzzy number into two parts: the position and fuzziness. The former is represented by the elements with membership value 1 and the latter by the entropy of the fuzzy number, both have crisp values. The conventional regression analysis is applied to find the relationship between the position (and entropy) of the fuzzy response variable and that of the fuzzy explanatory variables. Given a set of fuzzy explanatory variables, the position and entropy of the estimated fuzzy responses are calculated from the regression model. Via the one-to-one correspondence between a fuzzy number and its entropy, the estimated fuzzy response is obtained.
Keywords :
entropy; fuzzy set theory; least squares approximations; regression analysis; fuzzy explanatory variables; fuzzy number entropy; fuzzy regression analysis; fuzzy response variable; least-square method; membership value; one-to-one correspondence; regression model; Biological system modeling; Computational biology; Entropy; Fuzzy set theory; Fuzzy sets; Input variables; Linear regression; Quality management; Regression analysis;
Conference_Titel :
Intelligent Systems, 2004. Proceedings. 2004 2nd International IEEE Conference
Print_ISBN :
0-7803-8278-1
DOI :
10.1109/IS.2004.1344672
