Clustering large data sets based on data compression technique and weighted quality measures

Various algorithms have been proposed for clustering large data sets for the hard and fuzzy case, not as much work has been done for automatic clustering approaches in which the number of clusters is unknown for the user. These approaches need some measures, called validity function to evaluate the clustering result and to give to the user the optimal number of clusters. In order to obtain this number, three conditions are necessary: (1) a good compression technique for data reduction with limited memory allocated, (b) good measures for the evaluation of the goodness of clusters for varying number of clusters, and (c) a good cluster algorithm that can automatically produce the number of clusters and takes into account the used compression technique. In this paper, we propose new clustering approaches which deals with new compression technique based on quality measures.