Information Measures and some Distribution Approximations.

Abstract

The Fisher and Kullback- Liebler information measures were calculated from the approximation of a binomial distribution by both the Poisson and the normal distributions and are applied to the approximation of a Poisson distribution by a normal distribution. In this paper the concept of relative loss in information due to approximating the distribution of a random variable by that of another distribution of is introduced, and this concept is used to determine the value of the sample size for which the relative loss in information measure is less than a given level .