finite mixture model for SAS
in [SAS]
|
From: ajay ohri on 30 Nov 2009 06:05 Wensui you can try both mclust and mixtools in R as pre compiled macros/packages to make a single line code for making mix models in R The single line command would be mlcustbic(parameters...) I am not sure which SAS / license or Macro can make mix models its funny I was working on mix models but using parallel processing on Open MPI this morning for a project on a 1500 cluster. SAS pricing for multiple clusters is too high even for academics ( >100,000). Thats one reason the nearby Oakridge Lab with 100,000 processors or even UTenn's Kraken cant use SAS for HPC even for academic reasons. Regards, Ajay "If you have the guts, you will have the glory" ________________________________ From: Dale McLerran <stringplayer_2(a)YAHOO.COM> To: SAS-L(a)LISTSERV.UGA.EDU Sent: Sun, November 29, 2009 4:14:01 PM Subject: Re: finite mixture model for SAS Wensui, Can you elaborate on your question a little more? Dale --------------------------------------- Dale McLerran Fred Hutchinson Cancer Research Center mailto: dmclerra(a)NO_SPAMfhcrc.org Ph: (206) 667-2926 Fax: (206) 667-5977 --------------------------------------- --- On Sun, 11/29/09, Wensui Liu <liuwensui(a)GMAIL.COM> wrote: > From: Wensui Liu <liuwensui(a)GMAIL.COM> > Subject: Re: finite mixture model for SAS > To: SAS-L(a)LISTSERV.UGA.EDU > Date: Sunday, November 29, 2009, 6:58 AM > shouldn't we use aic / bic? it is > what i understand. > > On Sun, Nov 29, 2009 at 9:23 AM, Arthur Tabachneck <art297(a)netscape.net>wrote: > > > Take a look at: > > > > http://ca.wiley.com/WileyCDA/WileyTitle/productCd-0471006262.html > > http://www.listserv.uga.edu/cgi-bin/wa?A2=ind0503d&L=sas-l&P=27086 > > > > HTH, > > Art > > ----------- > > On Nov 29, 1:51 am, tamuheyii <t...(a)heyii.com> > wrote: > > > Is there any way to do finite mixture model in > SAS? I know SAS can > > > estimate the parameters once you know the number > of segments/classes > > > there (usually example is 2), but unlike other > programs, I can't find > > > anything about identification of the number of > segments/classes in the > > > data using SAS. They usually suggest doing > this with Mplus, Winbugs, > > > etc. But I am not familiar with any of that.. Any > suggestions? Thank > > > you! > > > > > > -- > ============================== > WenSui Liu > Blog : statcompute.spaces.live.com > Tough Times Never Last. But Tough People Do. - Robert > Schuller > ============================== >
From: Wensui Liu on 30 Nov 2009 19:02 Dale, It is not me who asked the question on selection of component numbers in FMM. I am just trying to answer the question. On page 202 of "finite mixture models" by McLachlan and Peel, it is suggested to use information criterions (aic / bic) to select number of components in a FMM. However, I really like to see how you answer this question, since your reply is the most educational on SAS-L. Have a nice holiday season! wensui On Sun, Nov 29, 2009 at 4:14 PM, Dale McLerran <stringplayer_2(a)yahoo.com>wrote: > Wensui, > > Can you elaborate on your question a little more? > > Dale > > --------------------------------------- > Dale McLerran > Fred Hutchinson Cancer Research Center > mailto: dmclerra(a)NO_SPAMfhcrc.org > Ph: (206) 667-2926 > Fax: (206) 667-5977 > --------------------------------------- > > > --- On Sun, 11/29/09, Wensui Liu <liuwensui(a)GMAIL.COM> wrote: > > > From: Wensui Liu <liuwensui(a)GMAIL.COM> > > Subject: Re: finite mixture model for SAS > > To: SAS-L(a)LISTSERV.UGA.EDU > > Date: Sunday, November 29, 2009, 6:58 AM > > shouldn't we use aic / bic? it is > > what i understand. > > > > On Sun, Nov 29, 2009 at 9:23 AM, Arthur Tabachneck <art297(a)netscape.net > >wrote: > > > > > Take a look at: > > > > > > http://ca.wiley.com/WileyCDA/WileyTitle/productCd-0471006262.html > > > http://www.listserv.uga.edu/cgi-bin/wa?A2=ind0503d&L=sas-l&P=27086 > > > > > > HTH, > > > Art > > > ----------- > > > On Nov 29, 1:51 am, tamuheyii <t...(a)heyii.com> > > wrote: > > > > Is there any way to do finite mixture model in > > SAS? I know SAS can > > > > estimate the parameters once you know the number > > of segments/classes > > > > there (usually example is 2), but unlike other > > programs, I can't find > > > > anything about identification of the number of > > segments/classes in the > > > > data using SAS. They usually suggest doing > > this with Mplus, Winbugs, > > > > etc. But I am not familiar with any of that.. Any > > suggestions? Thank > > > > you! > > > > > > > > > > > -- > > ============================== > > WenSui Liu > > Blog : statcompute.spaces.live.com > > Tough Times Never Last. But Tough People Do. - Robert > > Schuller > > ============================== > > > -- ============================== WenSui Liu Blog : statcompute.spaces.live.com Tough Times Never Last. But Tough People Do. - Robert Schuller ==============================
From: Wensui Liu on 30 Nov 2009 19:04 Ajay, I am not the one who asked the question. However, just FYI, FMM can be implemented in SAS. For details, you can search for Dale's related posts before. NLmixed is used to estimate a FMM. On Mon, Nov 30, 2009 at 6:05 AM, ajay ohri <ajayohri(a)yahoo.com> wrote: > Wensui > > you can try both mclust and mixtools in R as pre compiled macros/packages > to make a single line code for making mix models in R > > The single line command would be mlcustbic(parameters...) > > I am not sure which SAS / license or Macro can make mix models > > its funny I was working on mix models but using parallel processing on Open > MPI this morning for a project on a 1500 cluster. > > SAS pricing for multiple clusters is too high even for academics ( > >100,000). Thats one reason the nearby Oakridge Lab with 100,000 processors > or even UTenn's Kraken cant use SAS for HPC even for academic reasons. > > Regards, > > Ajay > > > > > > "If you have the guts, you will have the glory" > > > > > > > > ________________________________ > From: Dale McLerran <stringplayer_2(a)YAHOO.COM> > To: SAS-L(a)LISTSERV.UGA.EDU > Sent: Sun, November 29, 2009 4:14:01 PM > Subject: Re: finite mixture model for SAS > > Wensui, > > Can you elaborate on your question a little more? > > Dale > > --------------------------------------- > Dale McLerran > Fred Hutchinson Cancer Research Center > mailto: dmclerra(a)NO_SPAMfhcrc.org > Ph: (206) 667-2926 > Fax: (206) 667-5977 > --------------------------------------- > > > --- On Sun, 11/29/09, Wensui Liu <liuwensui(a)GMAIL.COM> wrote: > > > From: Wensui Liu <liuwensui(a)GMAIL.COM> > > Subject: Re: finite mixture model for SAS > > To: SAS-L(a)LISTSERV.UGA.EDU > > Date: Sunday, November 29, 2009, 6:58 AM > > shouldn't we use aic / bic? it is > > what i understand. > > > > On Sun, Nov 29, 2009 at 9:23 AM, Arthur Tabachneck <art297(a)netscape.net > >wrote: > > > > > Take a look at: > > > > > > http://ca.wiley.com/WileyCDA/WileyTitle/productCd-0471006262.html > > > http://www.listserv.uga.edu/cgi-bin/wa?A2=ind0503d&L=sas-l&P=27086 > > > > > > HTH, > > > Art > > > ----------- > > > On Nov 29, 1:51 am, tamuheyii <t...(a)heyii.com> > > wrote: > > > > Is there any way to do finite mixture model in > > SAS? I know SAS can > > > > estimate the parameters once you know the number > > of segments/classes > > > > there (usually example is 2), but unlike other > > programs, I can't find > > > > anything about identification of the number of > > segments/classes in the > > > > data using SAS. They usually suggest doing > > this with Mplus, Winbugs, > > > > etc. But I am not familiar with any of that.. Any > > suggestions? Thank > > > > you! > > > > > > > > > > > -- > > ============================== > > WenSui Liu > > Blog : statcompute.spaces.live.com > > Tough Times Never Last. But Tough People Do. - Robert > > Schuller > > ============================== > > > -- ============================== WenSui Liu Blog : statcompute.spaces.live.com Tough Times Never Last. But Tough People Do. - Robert Schuller ==============================
First
|
Prev
|
Pages: 1 2 Prev: Make Money From Home - How to Deal With Distractions Next: SAS/toolkit |