Building models with SAS Enterprise Miner, SAS Factory Miner, SAS Visual Data Mining and Machine Learning or just with programming

How can I decide the best number of clusters within 'Proc Fastclus'?

Reply
Frequent Contributor
Posts: 107

How can I decide the best number of clusters within 'Proc Fastclus'?

Hi, I have a question when working on cluster analysis.

 

How can I draw an Elbow Evaluation plot in determining the best number of clusters? Thanks!

 

 

proc fastclus
data=HOTEL_EMEA 
maxc=10
maxiter=10
out=clust;
var avg_pct;
run;
Super User
Posts: 9,662

Re: How can I decide the best number of clusters within 'Proc Fastclus'?

I would do Primary Component Analysis and get those two PC and polt them by proc sgplot ,and you could probably get N of clusters. 

Super Contributor
Posts: 426

Re: How can I decide the best number of clusters within 'Proc Fastclus'?

Ksharp,

 

Could you please give me one simple example? What do you mean by 'get those two PC' in your reply?

Super User
Posts: 9,662

Re: How can I decide the best number of clusters within 'Proc Fastclus'?

use proc princomp to get two primary component of your data. And plot them at X-Y axis .you can see the number of cluster you should split.

 

Here is an example:

 

http://blogs.sas.com/content/iml/2014/11/07/distribution-of-blood-types.html

Frequent Contributor
Posts: 107

Re: How can I decide the best number of clusters within 'Proc Fastclus'?

Cool!

Super Contributor
Posts: 426

Re: How can I decide the best number of clusters within 'Proc Fastclus'?

Can we cluster analysis only with continous varaibles?

Ask a Question
Discussion stats
  • 5 replies
  • 423 views
  • 0 likes
  • 3 in conversation