Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: RE: clustering question


From   "Data Analytics Corp." <walt@dataanalyticscorp.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: RE: clustering question
Date   Thu, 03 Dec 2009 15:42:26 -0500

Hi,

Unfortunately, a predecessor, who died last year, did just this - so now 
I'm expected to do it.   I suspected that I would have to run the whole 
all over again.  This is perhaps what he did.  Oh, well - I had to ask.

Thanks,

Walt


Nick Cox wrote:
> I can't see that the request makes any sense. 
>
> You might as well say "Last year, I calculated averages for the client,
> and the client was happy. Now the client has new data and wants new
> averages except that the old averages should be used in constructing
> them." You must choose: old or new or both. 
>
> All you can do is apply the same method to the new data, and that takes
> no cognisance of the old data. In fact, the new clustering might be some
> sort of check on the old, except that it may be too late for the client.
>
>
> There are variants on this such as combining old and new datasets. 
>
> It could be different, although not necessarily a better idea, for a
> method such as -pca-, which leaves pertinent results in its wake. 
>
> Nick 
> n.j.cox@durham.ac.uk 
>
> Data Analytics Corp.
>
> Last year, I developed a simple three-cluster solution for a client 
> based on survey data using the following:
>
>     cluster wardslinkage q16r6 q17r5 q17r7 q17aa1r1 q17ab1r1 q17ab1r7 
> q17ab1r9 vytldl, measure(Gower) name(clus01);
>     cluster dendrogram clus01, xlabel(, angle(90) labsize(*0.75)) 
> cutnumber(30) title("Wave 17 Cluster Analysis")
>         note("Notes:" "Based on Wards linkage" "062209_1 cluster.do");
>     cluster gen g = group(3), name(clus01);
>
> The client was happy.  Now this year, the client redid the survey but 
> wants to use the same cluster solution as last year, only with the data 
> from the new survey.  How do I use the previous solution with new data?
>
> I have both Stata 10 and 11 for Windows.
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>
>   


-- 
________________________

Walter R. Paczkowski, Ph.D.
Data Analytics Corp.
44 Hamilton Lane
Plainsboro, NJ 08536
________________________
(V) 609-936-8999
(F) 609-936-3733
walt@dataanalyticscorp.com
www.dataanalyticscorp.com


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index