International Journal of Computer Applications
Traditional k-means algorithm is well known for its clustering ability and efficiency on large amount of data sets. But this method is well suited for numeric values only and cannot be effectively used for categorical data sets. In this paper, the authors present modified k-means algorithms that can that can perform clustering very effectively on mixed data sets. The main intuition behind their proposed method is that all prototypes are the potential candidates at the root level. For the children of the root node, they can prune the candidate set by using simple geometrical constraints.