Design and Performance Analysis of Distributed Implementation of Apriori Algorithm in Grid Environment

Abstract

This paper presents the design and performance analysis of distributed implementation of Apriori algorithm in grid environment. Apriori algorithm is very important algorithm in data mining discipline that enables organizations to mine large amount of historical data that they gather over period of time and discover hidden patterns in that data. Data mining techniques enable organizations to analyze market trends and user behavior. If the data set to be mined is very large then varying the basic algorithm for execution in a distributed environment makes sense because distributed technologies generally offer performance benefits. Grids have gained wide popularity in executing a task in distributed fashion and offer performance benefits. So in this paper we have made an attempt to implement distributed version of basic Apriori algorithm in a grid environment. The Grid environment has been constructed using Globus® Toolkit. Experimental results show that distributed version offers performance benefits over basic version of Apriori algorithm and hence is a good implementation choice if the data to be mined is really large and distributed.