Hi; datamining question: ---------------------- my input data table is of size 1000-rows, 2000-columns. each column represents one features. one of the columns represent a categorical features with 900 levels. i would like to apply KNN / distance matrix on the table rows, while preserving right treatment for this categorical column and the other numerical columns. i understand that "gower" distance might fulfill this purpose if used, but i read that matrix/table need to be numerical in case "daisy" is being used for example. what is the right technique in such case? Thanks Rami _________________________________________________________________ Facebook. :ON:WL:en-US:SI_SB_facebook:082009 [[alternative HTML version deleted]]