MVLSC HomeIssue Contents

Componentwise Fuzzy Linear Regression Using Least Squares Estimation
Jin Hee Yoon and Seung Hoe Choi

This paper introduces a componentwise fuzzy linear regression model in order to construct a fuzzy relationship between fuzzy dependent and independent variables. We use the least squares method for an α-level set of observed fuzzy numbers in order to estimate the componentwise fuzzy regression model, which splits up response functions on mode and spreads of dependent variables in the fuzzy regression model.We also evaluate a mean value and a degree of fuzziness and closeness for predicted fuzzy numbers, and compare the accuracy of the proposed fuzzy regression model with other fuzzy regression models estimated by other authors.

Keywords: Componentwise fuzzy regression, least squares method, fuzziness, closeness, accuracy.

Full Text (IP)