Density estimation using entropy maximization for semi-continuous data

Semi-continuous data comes from a distribution that is mixture of the point mass at zero and continuous with support on positive real line. Such appear in several real-life situations like blind signal separation problems, modeling loss finance insurance, sales consumer goods, daily precipitation data, to name few. In this paper, we present novel algorithm estimate density function for semi-continuous using principle maximum entropy. Unlike existing methods literature, our provides consistent true entropy considers both discrete parts simultaneously. Using simulations, show produced by has significantly less bias compared methods. An application rainfall provided.