I'm working on proc prinqual to reduce the dimension of categorical variables (using opscore transformation). The scored dataset (output dataset from prinqual) was plugged into the proc varclus to get subset of statistically important variables. Is this a proper way to get statistically significant sub set of categorical variables? And, I could not figure out how the scores are assigned to each levels of categorical variables (transformed variables). I'm working with unsupervised dataset, do not have any target variables.
I really appreciate any comments or feedbacks.
Thanks!
_kamal