Minimum K-phi-divergence estimator

Abstract

In the present work, the problem of estimating parameters of statistical models for categorical data is analyzed. The minimum K-phi-divergence estimator is obtained minimizing the K-phi-divergence measure between the theoretical and the empirical probability vectors. Its asymptotic properties are obtained. Rom a simulation study, the conclusion is that our estimator emerges as an attractive alternative to the classical maximum likelihood estimator.