Abstract
A novel regression model with variable spreads fuzzy error is proposed for crisp explanatory and fuzzy response observations to tackle the spreads increasing problem. It can also cope with the situations of increasing, decreasing, constant or variable spreads. The coefficients of the model are estimated by using a least absolutes method. Then, based on minimizing the difference of membership values between the observed and estimated response variable, the fuzzy error terms are estimated. The results from comparative examples, based on some well-known data sets, show effectiveness of the proposed model.
