I want to know in case of proc fasclus how the distance is calculated and assigned to a particular cluster,I also want to know how the initial seed is chosen in case of proc fasclus and how the pseudo f stat is calculated?

[suk]

Can anyone of you create a sample data set and step by step explain how the initial seed is calculated in case of proc fasclus and how the distance is calculated from the respecttive cluster seed and according to the distance it is assigned to the cluster

and also how the pseudo f stat is calculated.

Thanks in advance.

[stat_sas]

Just to understand better, does your question relate to for proc fastclus?

[suk]

Yes I am talking about proc fastclus..Please reply

[SteveDenham]

Have you read the Introduction to Clustering Procedures in the SAS/STAT documentation?  Most of your questions can be answered there.

Steve Denham

[suk]

yes i read the SAS Documentation but still I am not able to understand.So I need help to understand it through an example.

Thanks in advance

[thebob]

Hi Suk,

I agree that the SAS documentation does not mention everything needed for how initial seeds are selected - I've started to write this process in open code but I'm struggling to get the first test right on datasets with more than 1,000 observations.

As for the PSF value, I have written code that will replicate this value. It didn't take long to calculate each component by starting with a small dataset (say, 20 observations) and only 2 variables to play around with. I suggest the following links:

• http://www.palgrave-journals.com/jibs/journal/v37/n4/fig_tab/8400206t2.html
• https://www.khanacademy.org/math/probability/statistics-inferential/anova/v/anova-2-calculating-ssw-...
• F-test - Wikipedia, the free encyclopedia