Date:         Tue, 2 Nov 2004 10:37:39 -0500
Reply-To:     Michael Bears <michael.bears@ANTHEM.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Michael Bears <michael.bears@ANTHEM.COM>
Subject:      GLMs and the Tweedie Distribution

Good morning - I was hoping that someone here had some experience in using
the tweedie distribution (as opposed to the gamma, neg binomial, etc.)
within PROC GENMOD. It does not appear to be offered as a standard
selection (true?), and I hope to learn how to use this distribution. The
response variable I am working with is positively skewed (gamma-ish) with a
probability mass at zero, which seemed like a good situation for the
Tweedie. Any thoughts would be very appreciated. Take care.