Performance based Efficient K-means Algorithm for Data Mining

Abstract

Today, we are witnessing enormous growth in data volume. Often, data is distributed or it can be in the form of streaming data.
Efficient clustering in this entire scenario becomes a very challenging problem. Our work is in the context of K-means clustering algorithm. Kmeans
clustering has been one of the popular clustering algorithms. It requires several passes on the entire dataset, which can make it very
expensive for large disk-resident datasets and also for streaming data. In view of this, a lot of work has been done on various approximate
versions of k-means, which require only one or a small number of passes on the entire dataset. In our work has developed a new algorithm for
very large data clustering which typically requires only one or a small number of passes on the entire dataset. The algorithm uses sampling to
create initial cluster centers, and then takes one or more passes over the entire dataset to adjust these cluster centers. We have implemented to
develop clustering algorithm for distributed data set. The main contribution of this paper is the implementation and evaluation of that algorithm.
Our experiments show that this framework can be very effective in clustering evolving streaming data.

Key words- Data Mining; Clustering; Distributed k-means Algorithm; Search Engine Technique.