A MapReduce based ISODATA algorithm

Abstract

Cluster analysis is a mathematical method that applied in various fields, such as biology, medicine, business and marketing. ISODATA algorithm is a Clustering algorithm that has been widely used. With the development of information technology, data set expanded dramatically which becomes a great challenge to the traditional algorithm. Parallel computing is one of the main methods to solve such problem. We have proposed a Parallel ISODATA algorithm based on MapReduce which is a famous distributed computing framework. Experiments show that it greatly improves the efficiency of the algorithm.