Prev: Looking for SAS job in Canada
Next: Using LIFEREG estimates for CDF/PDF
From: -bwg on 28 May 2010 10:22
Hi guys,

I have a rectangular matrix, approximately 2000x500, of response times
to test questions (actually the residuals after removing the row and
column effects). I want to permute the rows and columns of the matrix
so similar rows are adjacent and similar columns are adjacent. I don't
have access to SAS/IML (I almost wanted to call it Proc Matrix, talk
about old!). Does anyone have any suggestion for accomplishing my
goals?

I'm thinking of first clustering the rows and then the columns and
somehow using the ordering of the tree diagram, but a) I'm not sure
that's the best way and b) if it is the best way, I'm not sure which
clustering procedure and method to use. Having removed the main
effects, I have no reason to suspect the data are unidimensional.

Any suggestions, ideas or hints?

As always, thanks a lot for any help!

Barry Grau
 | 
Pages: 1
Prev: Looking for SAS job in Canada
Next: Using LIFEREG estimates for CDF/PDF