Actually this is not a large set of cases for modern
software.
My practice since the mid-70s, when I want a single nominal level
variable as a result and have continuous variables, is to consider
a cluster those cases that are put together by several methods of
agglomeration and varied contingency coefficients.
_A lot depends on exactly what you are trying to do._
Are you looking for a tree, a network, or a single nominal level
variable?
Is the clustering the final purpose, or do
you want to use the new variable in further analysis? If
so, what kind?
What is the nature of your cases? How were they selected?
What variables do you want to base the clustering on? What are
their levels of measurement? are they reasonably independent of
each other?
Art Kendall
Social Research Consultants
On 4/14/2012 3:53 AM, GauravSrivastava wrote:
Hi All,
I am performing cluster analysis for a large data set having 400-1000 cases.
Can anybody suggest which method and technique will be better for this.
Gaurav
--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Cluster-analysis-tp5640192p5640192.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Art Kendall
Social Research Consultants
Locked
