Study On New Approach For Effective Mining Association Rules From Huge Databases

Data mining is a logical process that is used to search through large amount of data in order to find useful patterns. The exponential growth of computer hardware and system software technology in the past three decades has led to large supplies of powerful and cost effective computer, data collection equipment and storage media. This technology provides a great boost to the database and information industry and makes a huge number of databases and information repositories available for transaction management information retrieval and data analysis. Extracting the association rule from this kinds of data has become vital and challenging problem in data mining.Association rule mining is a descriptive technique which can be defined as discovering meaningful patterns (itemsets tend to take place together in the transactions) from large collections of data.There is still nontrivial gap between general principles of association rules mining and its applications. Most of the previous studies were using Apriori-like algorithms to generate the association rules from the transactional database. This approach can suffer from two nontrivial costs:it needs to generate a huge number of candidate sets, and it may need to repeatedly scan the database and check a large set of candidates by pattern matching.The objective of this dissertation was to explore association rules and present a new model for mining the association rules from transactional database that can guarantee better performance than the priori-like models. The proposed model is using the integration of both the pattern growth ap-proach and apriori rule generation approach. Another aim was to apply the new model using real market basket dataset case study to assess its effectiveness. The knowledge obtained from the anal-ysis of the dataset using the proposed model can be used to improve the efficiency of a promotional campaign and a store layout.Several experimental applications performance were carried out on the collected data set and existing data set and the result of the study was that the new model outperformed the apriori-like models for both the dense data and the sparse data. This dissertation is mainly divided into five chapters which are summarized as follows:Chapter I provides the background and preliminaries on data mining. Chapter2presents formally the problem statement of frequent itemsets mining and shows the recent studies and research in the area of association rules mining. Chapter3Comparative study of frequent pattern growth approach. This chapter describes the main approaches used by our model. Chapter4presents a case study using our proposed model to find the association rules from the supermarket dataset. Chapter5pro-vides the evaluation of the proposed algorithm and Chapter5contains the conclusion and future work.