Derivation of ME estimator of a mixture of two normal distributions

Abstract

In this study, we obtain the parameters of a mixture density by using the principle of maximum equality which is based on Shannon's entropy measure. Since this procedure doesn't require constructing the likelihood function and give more accurate fitting values, it can be preferable instead of maximum likelihood method. Consequently, goodness of the derived estimator is shown by applying in order to obtain a mixture model under a real data application. The empirical study is done on the basis of MATLAB and R-packages.