Zero-one–inflated simplex regression models for the analysis of continuous proportion data

Abstract

Continuous data restricted in the closed unit interval [0,1] often appear in various fields. Neither the beta distribution nor the simplex distribution provides a satisfactory fitting for such data, since the densities of the two distributions are defined only in the open interval (0,1). To model continuous proportional data with excessive zeros and excessive ones, it is the first time that we propose a zero-one-inflated simplex (ZOIS) distribution, which can be viewed as a mixture of the Bernoulli distribution and the simplex distribution. Besides, we introduce a new minorization–maximization (MM) algorithm to calculate the maximum likelihood estimates (MLEs) of parameters in the simplex distribution without covariates. Likelihood-based inference methods for the ZOIS regression model are also provided. Some simulation studies are performed and the hospital stay data of Barcelona in 1988 and 1990 are analyzed to illustrate the proposed methods. The comparison between the ZOIS model and the zero-one-inflated beta (ZOIB) model is also presented.