[R-sig-Geo] Fitting Zero-Inflated Spatial Model in R

[Lauria, Valentina]

Hi Arnold,

To fit zero inflated model you can either use zero-inflated Poisson, zero-inflated negative binomial regression, and hurdle models from the pscl package in R.

However I would advise to spend a bit of time to see what type of zero you have in your data.
In particular the zero-inflation model is a mixture model of two components:

1 - a count component described by Poisson, NB, and 2- a zero mass component (i.e., zero with probability 1).

The observed zeros in your data can come from both sources: either they are "random" zeros from component (1) or "excess" zeros from component (2). There is a paper from Mei-Chen Hu et al. 2011 " Zero-inflated and Hurdle Models of Count Data with Extra Zeros:
Examples from an HIV-Risk Reduction Intervention Trial" where this is well explained.

Hope this helps,
Valentina


Dr. Valentina Lauria
Postdoctoral Researcher Associate 
Room 118, Ryan Institute
University Road
National University of Ireland, Galway
Galway
Ireland