ZIGP-package {ZIGP}R Documentation

Zero-Inflated Generalized Poisson (ZIGP) Models

Description

Adapts GP, ZIP and ZIGP models by Maximum Likelihood estimation. Regression is allowed not only on the mean but also on overdispersion and zero-inflation level. Therefore, three design matrices are allowed.

Details

Package: ZIGP
Type: Package
Version: 2.1
Date: 2008-03-24
License: GPL (>= 3)

Distribution functions in R notation are 'dzigp', 'pzigp', 'qzigp', 'rzigp'.

The most important function is 'wald.test'. This function can be fed with design matrices for the mean, overdispersion and zero-inflation level. It returns an summary-like overview of the covariates together with significance statistics referring to the Wald test. It can be used for the sequential elimination of non-significant effects.

Other useful functions are 'loglikelihood.zigp', which evaluates the loglikelihood function on the given parameter value. Scores can be calculated using 'gradient', the Fisher Information matrix using 'FM'.

Author(s)

Vinzenz Erhardt

Maintainer: Vinzenz Erhardt <erhardt@ma.tum.de>

References

Czado, C., Erhardt, V., Min, A., Wagner, S. (2007) Zero-inflated generalized Poisson models with regression effects on the mean, dispersion and zero-inflation level applied to patent outsourcing rates. Statistical Modelling 7 (2), 125-153.

Masterthesis in German: Erhardt, Vinzenz. Verallgemeinerte Poisson und Nullenueberschuss- Regressionsmodelle mit regressiertem Erwartungswert, Dispersions- und Nullenüberschuß-Parameter und eine Anwendung zur Patentmodellierung. ("http://www-m4.ma.tum.de/Diplarb/"), 2006


[Package ZIGP version 2.1 Index]