Date:         Fri, 15 Jul 2005 16:58:59 -0400
Reply-To:     "Feinstein, Zachary" <ZFeinstein@HarrisInteractive.com>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Feinstein, Zachary" <ZFeinstein@HarrisInteractive.com>
Subject:      FW: Cluster Analysis Seems to Have Too Many Variables
Content-Type: text/plain; charset="us-ascii"

I am sorry about the oversight below.  I meant to say 1,619 respondents.
I only have four standardized variables going into the clustering.  My
apologies.

 

Zachary

 

________________________________

From: Feinstein, Zachary 
Sent: Friday, July 15, 2005 1:27 PM
To: SPSSX-L@LISTSERV.UGA.EDU
Subject: Cluster Analysis Seems to Have Too Many Variables

 

I have 1,619 variables.  I am starting by doing a hierarchical cluster
analysis.  From there I will use the centroids as seeds for the quick
cluster.  The following code works with no problem if I bring the number
of cases down to 500.  Is there a way for me to tell SPSS (v13) to set
aside more memory?  I have found SET WORKSPACE but I'm at a bit of a
loss in figuring how much of a change I should make without compromising
other stuff I do in SPSS.

 

PROXIMITIES

 zcv1 TO zcv4

 /MATRIX OUT ("C:\temp2\spssclus.tmp")

 /VIEW = CASE

 /MEASURE = SEUCLID

 /PRINT NONE

 /STANDARDIZE = NONE.

 

SORT CASES BY id.

EXECUTE.

 

CLUSTER

 /MATRIX IN ("C:\temp2\spssclus.tmp")

 /METHOD WARD

 /PRINT SCHEDULE CLUSTER(3, 7)

 /PRINT DISTANCE

 /PLOT NONE

 /SAVE CLUSTER(3, 7).