Journal of applied mathematics & informatics
- Volume 18 Issue 1_2
- /
- Pages.591-601
- /
- 2005
- /
- 2734-1194(pISSN)
- /
- 2234-8417(eISSN)
ROBUST FUZZY LINEAR REGRESSION BASED ON M-ESTIMATORS
Abstract
The results of fuzzy linear regression are very sensitive to irregular data. When this points exist in a set of data, a fuzzy linear regression model can be incorrectly interpreted. The purpose of this paper is to detect irregular data and to propose robust fuzzy linear regression based on M-estimators with triangular fuzzy regression coefficients for crisp input-output data. Numerical example shows that irregular data can be detected by using the residuals based on M-estimators, and the proposed robust fuzzy linear regression is very resistant to this points.
Keywords