The KCLUS Procedure

For clustering inputs that include only interval or only nominal variables, the KCLUS procedure implements the k-means or k-modes clustering algorithm, respectively. For clustering inputs that include both interval and nominal variables, the KCLUS procedure implements the k-prototypes clustering algorithm (Huang 1997). The k-prototypes clustering algorithm uses a mixed distance that combines the distance contributed by the interval inputs and the distance contributed by the nominal inputs. This mixed distance is characterized with a weight parameter, , to balance the scales of the interval and nominal distances.