Row-Column Interaction Models for Zero-Inflated Poisson Count Data in Agricultural Trial

Abstract

Many zero observations makes some difficulties and fatal consequence in Poisson modeling and its interpretation. We consider to facilitates the analysis of two-way tables of count with many zero observations in agricultural trial. For example, in counting the pest or disease in plants. Plants that have no sign of attack, can occur because of two things, it could be resistant, or simply there is no spore disease (no endemics) or no pest attack. This is the difference between inevitable structural zero or sampling zero that is occurring according to a random process. This paper describes a statistical framework and software for fitting row-column interaction models (RCIMs) to two-way table of count with some Zero observations. RCIMs apply some link function to the mean of a cell equaling a row effect plus a column effect plus an interaction term is modeled as a reduced-rank regression with rank of 2, then will be visualized by biplot. Therefore its potentially to be develop become AMMI models that accommodate ZIP count.