Parameter Estimation in Gamma Mixture Model using Normal-based Approximation

Abstract

Gamma mixture models have wide applications in hydrology, finance and reliability. Parameter estimation in this class of models is a challenging task owing to the complexity associated with the model structure. In this paper, a novel approach is proposed to estimate the parameters of Gamma mixture models using Wilson-Hilferty normal-based approximation method. The proposed methodology uses a popular clustering algorithm for Gaussian mixtures namely, MCLUST and a confidence interval based search approach to obtain the estimates. The methodology is implemented on simulated as well as real-life datasets and its performance is compared with gammamixEM() function available in R.