ADJUSTED EXPLAINED VARIATION MEASURE FOR ZERO-INFLATED POISSON REGRESSION MODEL

ADJUSTED EXPLAINED VARIATION MEASURE FOR ZERO-INFLATED POISSON REGRESSION MODEL

Y. Al-Taweel, Z. Algamal

[PDF]

Abstract

Explained variation measures are very popular in regression models. They explain the amount of variation in the dependent variable by the explanatory variables. In some situations, a small sample size is available in comparison to the number of explanatory variables. In this case, explained variation measures may become significantly inflated. In order to tackle this problem, adjusted explained variation measures are used. In this work, we propose adjusted explained variation measures depend on the deviance for zero-inflated Poisson regression model (ZIPRM). We compare the performance of the suggested measures with that of the unadjusted explained variation measures for the ZIPRM using a Monte Carlo simulation experiment and real datasets.

Keywords

Poisson distribution, zero-inflated Poisson regression model, adjusted R2 measure, deviance.