Implementation and Comparative Study of Improved Apriori Algorithm For Association Pattern Mining
The data mining includes different kinds of data models for analysing the data. The data mining based analysis
leads to produce the outcomes according to the employed algorithms over similar kind of data. Thus according to the
databases and their mining based outcomes data mining algorithms can be classified in association rule mining algorithm,
classification algorithm or clustering methods. In literature a number of different techniques and algorithms are available by
which the association rules are mined for different purposes. Among them Apriori algorithm is one of most popular
technique of association rule mining. The Apriori algorithm is basically used for transactional pattern analysis using the
frequent pattern evaluation of target item sets. Therefore to execute the process, algorithm generates the candidate sets for
association pattern analysis.
In this presented work first the implementation of Apriori algorithm is performed and then to reduce the time and space
complexity a new technique using the Apriori algorithm is performed for finding the association rules. In the proposed rule
mining technique the candidate generation is limited by pre-analysis of itemsets during candidate set generation process. Due
to this less number of candidates and high quality rules are formed. The implementation of Apriori algorithm is performed
using JAVA technology. Additionally the performance in terms of space and time complexity is measured. According to
comparative performance study proposed Apriori algorithm provides better and less number of rules in efficient manner.
Keywords- Transactional database, Association pattern, Apriori algorithm, implementation, candidate set regulation.