2 PART MODEL
in [SAS]
|
From: raj on 22 Nov 2009 21:01 Hi, I am trying to do a 2 part model for health care cost analysis. My response variable is total cost and my predictor variables are categorical (binary) as well as continuous variables. I am comparing total costs of 2 treatment groups. I know 2 part model involves proc logistic and then proc genmod with gamma dist and log link. But i need to know some things with regard to the methodology and sas code. My specific questions are: 1) How do i get the average predicted probability of each TREATMENT group 2) How do i output adjusted means into a dataset in proc genmod. If anybody can provide me with a sas code, i would be grateful to them. Thanks for your help. KRB
From: Dale McLerran on 24 Nov 2009 17:40 Raj, Your two part model is what has been referred to elsewhere as a zero-inflated gamma distribution. With the zero-inflated gamma distribution, both parts of the model are estimated in a single procedure. I have posted previously how to fit a zero-inflated gamma distribution using the NLMIXED procedure. See my post at http://listserv.uga.edu/cgi-bin/wa?A2=ind0805A&L=sas-l&P=R20779 The specific questions you have below are addressed to some extent in that posting. If you have questions after reviewing that post, write back with a little more detail about your specific problem. Dale --------------------------------------- Dale McLerran Fred Hutchinson Cancer Research Center mailto: dmclerra(a)NO_SPAMfhcrc.org Ph: (206) 667-2926 Fax: (206) 667-5977 --------------------------------------- --- On Sun, 11/22/09, raj <rkamalakar73(a)GMAIL.COM> wrote: > From: raj <rkamalakar73(a)GMAIL.COM> > Subject: 2 PART MODEL > To: SAS-L(a)LISTSERV.UGA.EDU > Date: Sunday, November 22, 2009, 6:01 PM > Hi, > > I am trying to do a 2 part model for health care cost > analysis. > My response variable is total cost and my predictor > variables are > categorical (binary) as well as continuous variables. > I am comparing > total costs of 2 treatment groups. > > I know 2 part model involves proc logistic and then proc > genmod with > gamma dist and log link. But i need to know some things > with regard to > the methodology and sas code. > > My specific questions are: > 1) How do i get the average predicted probability of each > TREATMENT > group > 2) How do i output adjusted means into a dataset in proc > genmod. > > If anybody can provide me with a sas code, i would be > grateful to > them. Thanks for your help. > > KRB >
|
Pages: 1 Prev: fixed effect in Poisson regression of panel data Next: Logistic regression |