Generalized Maximum Entropy Estimation in Measurement Error Models

Abstract

Generalized maximum entropy approach is proposed by Golan et al. [6] to overcome the problem of collinearity and ill-conditioning among the explanatory variables in ordinary linear models. In this paper we extend this approach to the functional measurement error models. We derive generalized maximum entropy and generalized cross entropy estimate of parameters in this context. We illustrate our idea via an example from real data. Finally, based on this data set we perform a parametric bootstrap simulation study to compare entropy estimators with ridge and maximum likelihood estimators.