Re: [R] RE

From: Spencer Graves <spencer.graves_at_pdf.com>
Date: Wed 09 Aug 2006 - 18:22:45 EST

          I've never used the 'bayesm' package, but your question sounded interesting and I've not seen a reply, so I decided to look at it. Unfortunately, I couldn't make sense of your example in the time available. Instead, I will offer a few more general questions / suggests that I hope might help you.

  1. Have you worked through the example in the 'rmnlIndepMetrop' help file?
  2. Have you worked through the documentation in '~library\basesm\doc' subdirectory of your installed version of R?
  3. Have you tried the web site for the companion book, cited in the 'References' to the 'rmnlIndepMetrop' help file?

          If you've tried all these things and you would still like help from this listserve, I suggest you submit another post that is more consistent with the suggestions in the posting guide 'www.R-project.org/posting-guide.html'. This includes (a) using an informative subject (b) including the email address of the maintainer for the 'bayesm' package, listed in help(package='bayesm') in the distribution, and (c) providing commented, minimal, self-contained, reproducible code.

          The example you provided was helpful be inadequate for someone like me to diagnose the problem in a reasonable period of time. I suggest you include a call to 'set.seed' before your random number generation, and including attempted function calls to 'llmnl' and 'rmnlIndepMetrop' with comments explaining your confusion.

	  I know this doesn't answer your question, but I hope it helps.
	  Spencer Graves

nmi13 wrote:
> Hi any,
>
> Can some please detail me the createX command in bayesm package?
>
> To make things easy for you to help me, let me put forward my problem
>
> Suppose I have 3 covariates (say X matrix) and my Y has 3 categories say
> (1,2,3). Now from the CreateX I understand that the data matrix say 'Xa' must
> be of dimension n* (naxp), where 'na' is the number of variables and 'p' is
> the number of categories that Y has and 'n' is the number of observations. Now
> the Xa matrix will have 9 columns if I give na=3 and p=3, and 6 columns if I
> give na =2 and p=3. I understand this part. In order to create Xa with a
> dimension of n*9 or n*6 we have to create Xa as cbind(Xa,-Xa) and now when I
> get the design matrix say XD then I have 5 or 4 variables, which will be same
> as the beta matrix that I intend to get, I get this, but my question is when
> I get the XD matrix as explained below the covariate matrix (X) which
> initially had 3 columns now changed to a 9 rows and one column and two
> additional variables X4 and X5 as explained belwo which I guess for the beta,
> what is the role of these variables in the 'rmnlIndepMetrop'analysis.
> example
> x1<-runif(3,-1,1)
> x2<-runif(3,0,1)
> x3<-runif(3,10,50)
> X<-cbind(x1,x2,x3)
> X
> x1 x2 x3
> [1,] -0.9701396 0.4084203 41.31097
> [2,] 0.3844539 0.4791997 36.85861
> [3,] 0.2732056 0.5433642 13.14610
>
> Xa<-cbind(X,-X)
> XD<-createX(p=3,na=2,nd=NULL,Xa=Xa,Xd=NULL)
> XD
> [,1] [,2] [,3] [,4]
> [1,] 1 0 -0.9701396 0.9701396
> [2,] 0 1 0.4084203 -0.4084203
> [3,] 0 0 41.3109655 -41.3109655
> [4,] 1 0 0.3844539 -0.3844539
> [5,] 0 1 0.4791997 -0.4791997
> [6,] 0 0 36.8586070 -36.8586070
> [7,] 1 0 0.2732056 -0.2732056
> [8,] 0 1 0.5433642 -0.5433642
> [9,] 0 0 13.1461040 -13.1461040
> Xa<-cbind(X,-X,X^2) (is this a correct way)
> XD<-createX(p=3,na=3,nd=NULL,Xa=Xa,Xd=NULL)
> XD
> [,1] [,2] [,3] [,4] [,5]
> [1,] 1 0 -0.9701396 0.9701396 0.9411709
> [2,] 0 1 0.4084203 -0.4084203 0.1668071
> [3,] 0 0 41.3109655 -41.3109655 1706.5958746
> [4,] 1 0 0.3844539 -0.3844539 0.1478048
> [5,] 0 1 0.4791997 -0.4791997 0.2296324
> [6,] 0 0 36.8586070 -36.8586070 1358.5569127
> [7,] 1 0 0.2732056 -0.2732056 0.0746413
> [8,] 0 1 0.5433642 -0.5433642 0.2952447
> [9,] 0 0 13.1461040 -13.1461040 172.8200512
>
> In the above example my X matrix as you can see has 3 columns with 3
> observations, which now in XD are 9 observations in 3rd column, I don't know
> how col 4 and col 5 of XD play a role in computing the llmnl and
> rmnlIndepMetrop.
>
> Thanks for all your help and time.
>
> Regards,
> Murthy.
>
> ______________________________________________
> R-help@stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.



R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. Received on Wed Aug 09 18:27:51 2006

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.1.8, at Wed 09 Aug 2006 - 20:22:51 EST.

Mailing list information is available at https://stat.ethz.ch/mailman/listinfo/r-help. Please read the posting guide before posting to the list.