Estimating sample size to approximate some sampling distributions by information measures (original) (raw)

Estimating sample size to approximate some sampling distributions by information measures

B-entropy measure, Fisher information measures and Akaike information criterion are considered as three different types of information measures, entropy, parametric and statistical measures respectively. The main objective of this paper is to estimate the optimal sample size under which a random variable belonging to Gamma or Poisson distribution can be approximated by a random variable following the normal distribution in the sense of the central limit theorem, based on the concept of the percentage relative error in information due to approximation. The idea is to determining the sample size for which the percentage relative error in information measure is less than a given accuracy level 100𝜖% for small 𝜖 > 0.