Package VGAM for R has dzipois, the density function of zero
inflated poisson, and examples. If n is sample size and r is the
number of zeros, r/n should be a good initial estimator of phi, the
extra probability of zero, and coefficients from the usual poisson
regression should be good enough as starting values for
the regression parameters.
Kjetil Halvorsen
Peter Flom wrote:
>Hello
>
>I've been reading up about Zero Inflated Poisson and Zero Inflated
>Negative Binomial regression models. I am going to implement these in
>R, using a package developed by Lindsey.
>
>One thing I haven't seen is material on choice of initial estimates for
>the parameters. Does anyone know of a good refreence on this?
>
>Thanks in advance
>
>Peter
>
>Peter L. Flom, PhD
>Assistant Director, Statistics and Data Analysis Core
>Center for Drug Use and HIV Research
>National Development and Research Institutes
>71 W. 23rd St
>www.peterflom.com
>New York, NY 10010
>(212) 845-4485 (voice)
>(917) 438-0894 (fax)
>
>
>
>
|