10-13-2013 06:31 PM
I am new to cluster analysis and also new to SAS statistical procedures. Please help me understand the convergence criterion in fastclus procedure.
In the SAS documentation, in the PROC FASTCLUS example (Fisher Iris Data) below, the convergence criterion is shown to be satisfied:
The value of Converge is given as 0.02. The LEAST option is not specified and the Least Squares Criterion is probably used.
I want to know how the values in the "Criterion" column in the "Iteration History" are calculated. How is this related to the converge value 0.02, the relative change in cluster seeds, the minimum distance between initial seeds, etc. Please clarify.