Title :
A new resolution of fuzzy regression analysis
Author :
Alex, Rajan ; Wang, P.Z.
Author_Institution :
Dept. of Math, West Texas A&M Univ., Canyon, TX, USA
Abstract :
In this paper, we describe a method for fuzzy regression analysis. Earlier work on regression analysis using fuzzy models made heavy emphasis on the type of membership function used in fuzzifying the predictor and the response variables. Our proposed solution of fuzzy regression analysis does not depend on the type of fuzzy function used to fuzzify the given data. The fuzzy regression analysis is essentially to answer the question: What is the predicted value of the response variable y for a given set of n values of the predictor variables? The first step is to fuzzify each value of the involved variables to a fuzzy number. Then establish a fuzzy relationship between the independent variables and the dependent variable without taking into consideration the particular type of fuzzy number that has been assigned to the data. Through an example we illustrate the application of the fuzzy modeling
Keywords :
fuzzy set theory; statistical analysis; fuzzy modeling; fuzzy number; fuzzy regression analysis; predictor fuzzification; response variables fuzzification; Feeds; Fuzzy sets; Linear regression; Mathematics; Multi-layer neural network; Neural networks; Predictive models; Regression analysis;
Conference_Titel :
Systems, Man, and Cybernetics, 1998. 1998 IEEE International Conference on
Conference_Location :
San Diego, CA
Print_ISBN :
0-7803-4778-1
DOI :
10.1109/ICSMC.1998.728194