EL CUBRIMIENTO DE UNA MUESTRA: ESTIMACIÓN Y PREDICCIÓN

Abstract

The problem of estimating or predicting the number of classes in population U is present in many applications and it outlines several queries to statistics. Many of them have been not solved yet. Good (1953) and Esty (1986) developed Central Limit Theorems. Their hypothesis are somewhat strong. In this paper we propose the use of a regression type superpopulation model. Its application in this case leads to a Generalized Linear Model as a describer of the behaviour of the coverage. It does not depend on the equiprobability hypothesis to guarantee the asymptotic normal distribution of the predictor. To analyse the behaviour of the diverse alternatives that characterize the coverage of the sample we use data of two investigations. One of them is the data obtained from the study of the infestation of cane fields by a plague in Cuba. The other one consists is a study of forest biodiversity in Mexico