K-means algorithm based Clustering for Big data

Clustering is a data mining technique used to place data elements into related groups without advance knowledge of the group definition. Clustering is a pro-cess of partitioning a set of data in a set of meaningful sub-classes, called cluster. In this paper, we propose to give a review of the most used clustering methods. First, we give an introduction about clustering methods, how they work and their main challenges. Second, we present the clustering methods with some comparisons including mainly the classical partitioning clustering methods like well-known k-means algorithms, Gaussian Mixture Modals and their variants, the classical hierarchical clustering methods. Clustering algorithms can be categorized into partition-based algorithms, hierarchical-based algorithms, density-based algorithms and grid-based algorithms. Partitioning clustering algorithm splits the data points into k partition, where each partition represents a cluster. Hierarchical clustering is a technique of clustering which divide the similar dataset by constructing a hierarchy of clusters. Density based algorithms and the cluster according to the regions which grow with high density. It is the one-scan algorithms. Grid Density based algorithm uses the multi resolution grid data structure and use dense grids to form clusters. Its main distinctiveness is the fastest processing time. In this survey paper, an analysis of clustering and its different techniques in data mining is done.