Events and Meetings of Italian Statistical Society, Advances in Latent Variables - Methods, Models and Applications

Font Size: 
A Generalized Maximum Entropy (GME) approach to crisp-input/fuzzy-output regression model
Antonio Calcagnì

Last modified: 2013-06-14

Abstract


In this short-paper we describe the application of Generalized Maximum Entropy Method of Estimation on crisp-input/fuzzy-output regression model. In order to highlight some interesting features provided by this method, we carried out a Monte Carlo experiment in which both models were tested by varying multicollinearity in the design matrix and by considering two levels of sample sizes. Next, the performances of the two methods were evaluate in terms of standard errors of the regression coefficient and RMSEs for the fuzzy dependent variables.

Full Text: PDF