Dear All,
I've been recently been implementing some k-modes/k-medoids type procedures in R, such as k-prototypes, and the daisy package.
The output is fine. Yet, I wondered if anyone was aware of clustering packages for categorical data where the range of data is small such as [1,2,3].
My suspicion with such limited range is that the mode/medoid might not discriminate as accurately as possible. However, this is just a speculative question.
Would love to hear some thoughts.