On Tue, 28 Sep 1999, Pierre Delfosse wrote:
>
> Dear S+ users,
>
> I would like to build confidence intervals for a discrete Poisson
> variable whose mean follows a normal distribution with mean>0 and CV<<1
> (at least approximately, given the positivity constraint).
> As this mean can take small values, the usual normal approximation to
> the Poisson variable cannot be accepted.
>
> Thank you for any help (including a pointer to a good reference).
>
A more natural mixture model for the Poisson is the Gamma mixture
(Poisson mean follows a Gamma distribution). This will automatically
satisfy the positivity constraint... A Gamma mixture of Poissons
has a Negative Binomial distribution.
albyn
------------------------------------------------------------------
http://www.reed.edu/~jones Albyn Jones jones@reed.edu
Reed College, Portland OR 97202 (503)-771-1112 x7418
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
|