Standard Article

Zero-Inflated Data

General

  1. Wanzhu Tu

Published Online: 15 SEP 2006

DOI: 10.1002/9780470057339.vaz000g

Encyclopedia of Environmetrics

Encyclopedia of Environmetrics

How to Cite

Tu, W. 2006. Zero-Inflated Data. Encyclopedia of Environmetrics. 6.

Author Information

  1. Regenstrief Institute, Indianapolis, IN, USA

Publication History

  1. Published Online: 15 SEP 2006

Abstract

As its name suggests, zero inflation indicates that a dataset contains an excessive number of zeros. The word inflation is used to emphasize that the probability mass at the point zero exceeds that allowed under a standard parametric family of distributions. Zero-inflated data are abundant in many disciplines, including the environmental sciences. If not properly modeled, the presence of excess zeros can invalidate the distributional assumptions of the analysis, jeopardizing the integrity of the scientific inferences. The zeros can also create many computational difficulties. Recently, a fair amount of interest has concentrated on the analysis of zero-inflated Poisson (ZIP) counts. Strictly speaking, however, the data need not follow a Poisson distribution in order to be zero inflated. In this article, a broader definition of the subject is taken; zero inflation is considered as a practical data phenomenon that can be associated with any parametric family of distributions.