Generally Altered, Inflated and Truncated Poisson and Binomial Distributions
Thomas Yee1 and Chenchen Ma2
1. Department of Statistics, University of Auckland
2. Department of Mathematics, Zhejiang University, China
The zero-altered (hurdle), zero-inflated and zero-truncated distributions are very well-known to statistical practitioners for handling surplus and absences of 0s in responses. This talk sketches early work on generalizing these to other values of the support so that 0 is not necessarily the only special value. We focus on the Poisson and binomial models, however the work is easily extended to other count distributions. Software implementing some of the methods will also be described; they include several new family functions in the VGAM R package, as well as d- and p- and q-type functions. Some applications will also be described. This is also work with Theodora Jin.