Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: zero-inflated Poisson distribution fitting for correlated count data


From   "Stas Kolenikov" <skolenik@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: zero-inflated Poisson distribution fitting for correlated count data
Date   Sun, 11 Mar 2007 22:41:44 -0500

There is -xtnbreg-, look it up and see if it does the job you need.

If it does not, I would expect that with trial and a lot of errors,
you can squeeze something of a kind from -gllamm- (which is a terrific
piece of software for lots of modeling issues). The basic outline is
to have a 2-point latent variable to model the mixture giving rise to
zero-inflated Poisson; and then use Poisson link for one part of the
mixture, and binomial link with the other. I am thinking aloud there,
I don't know if it is really feasible, so look up the book somebody
just mentioned by Sophia Rabe-Hesketh and Anders Skrondal.

On 3/11/07, Dr. Stephen Rothenberg <srothenberg@mda.cinvestav.mx> wrote:
I'm trying to determine the parameters of a distribution of a time-series of
yearly number of disease cases over a 33 year period.  The series has many
zero values, and the variance is much larger than the mean (var=13.6,
mean=1.6).  Since the data is taken from one city over a number of years,
serial correlation may be expected.

My goal is to determine the probability of x number or greater cases in any
given year, given the entire series.

I'm looking for something like -nbfit- for correlated data.  Or -nbreg-
(with independent variables, like year) for correlated data.


--
Stas Kolenikov
http://stas.kolenikov.name
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index