An improved k-medoids clustering algorithm

In this paper, we present an improved k-medoids clustering algorithm based on CF-Tree. The algorithm based on the clustering features of BIRCH algorithm, the concept of k-medoids algorithm has been improved. We preserve all the training sample data in an CF-Tree, then use k-medoids method to cluster the CF in leaf nodes of CF-Tree. Eventually, we can get k clusters from the root of the CF-Tree. This algorithm improves obviously the drawbacks of the k-medoids algorithm, such as the time complexity, scalability on large dataset, and can´t find the clusters of sizes different very much and the convex shapes. Experiments show that this algorithm enhances the quality and scalability of clustering.