k means clustering

[jinc]

I have a set of data on the credit history of customers and I have to create clusters based on unique customers. The dataset has both character variable and continuos variables.

Can you please tell the syntax for k means clustering in such a case and the method to be used.

[Ksharp]

I could think two ways.

1) using PROC DISTANCE to get the mixed data distance matrix,
then feed it into PROC CLUSTER.

2) using PROC PRINQUAL to get the first two principal component,
and plot obs by these two principal component, you can see which one is closer to which one.