Data Decomposition Technique Proposed for Candidate Itemsets Generation of Association Rule Mining Algorithms on Heterogeneous Cluster
Rakhi Garg1, P. K. Mishra2

1Dr. Rakhi Garg, Department of Computer Science, MMV, Banaras Hindu University, Varanasi (U.P), India.
2Prof. P. K. Mishra, Department of Computer Science, Banaras Hindu University, Varanasi (U.P), India.
Manuscript received on 12 March 2013 | Revised Manuscript received on 21 March 2013 | Manuscript Published on 30 March 2013 | PP: 72-75 | Volume-2 Issue-4, March 2013 | Retrieval Number: D0526032413/13©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Among various data mining task, association rule mining (ARM) is the major technique which is widely used in retail marketing, bioinformatics, website navigation analysis etc. It finds correlations among items in a given data sets and establishes an association between two non overlapping sets of frequently occurring values in a database. Various sequential and parallel ARM algorithms have been developed that differs in data layout, search technique, data structure, the number of database scans used and the system on which it is developed i.e. homogeneous or heterogeneous systems. This paper mainly put emphasis in the need of a candidate based ARM algorithm for heterogeneous PC cluster that reduces the database scans and time complexity. It also describe the design and functioning of the heterogeneous PC cluster i.e. MPICH2 and the data decomposition technique applied for candidate itemsets generation that plays important role in balancing workload as well as enhancing the performance of the algorithm on MPICH2 heterogeneous PC cluster.
Keywords: Association Rule Mining, Candidate 1-Itemsets, Data Mining, Heterogeneous Cluster.

Scope of the Article: Clustering