A mixture model approach to constructing a bivariate generalized exponential distribution and its application for Kepler data

DOI: 10.5176/2251-1911_CMCGS18.34

Authors: Yen-Chang Chang, Jen-Hao Chen, Wen-Liang Hung

Abstract:Mixture models are statistical tools, with many applications including data mining, pattern recognition, machine learning and statistical analysis. This paper proposes the use of mixture models to construct a new bivariate generalized exponential (BVGE) distribution. One important application of the proposed BVGE distribution is to fit the period-radius probability density function for Kepler planet single transits and estimate the occurrence rate of Earth analogs around GK dwarfs.

Keywords: EM algorithm; Finite mixture models; Fisher information matrix; Generalized exponential distribution; Kepler exoplanets