Clustering transactional data

Giannotti F, Gozzi C, Manco G

Transactional data clustering  K-Means Algoritm 

This paper presents a partitioned method which crosses the limitations of traditional approaches to clustering of transactional data. A modification of the stanard K-Means algorithm is presented, which has a good scalability on the number of objects and attributes, but can only work with numeric vectors of fixed length.