Association Rule Mining Algorithm Based On Rough Set

With rapid development of information age and widely application of database, there is aphenomenon on data inflation. People desire to mine the relation of these data, therefore to getuseful information, which is a very difficult task. Association rule mining comes out at thehistoric moment, of which the goal is to find out the relationship between the data and come intobeing rules. Apriori is the most classical algorithm in the field of association rule mining. Thereexists many deficiencies on Apriori. Although there are many improved algorithm, Some of thesealgorithms still can generate candidate item sets, and some can only apply to small and mediumscale of data sets.In recently years,"Inflation" of data is more and more serious, and people don’t know whatto do. The technology of association rules mining can not mine all the useful information fromhuge amounts of data effectively. Rough set theory comes out at the historic moment, and ismoving to the field of association rules mining gradually, which are mainly reflected in threeaspects:Firstly, in the data preprocessing stage, rough set can fill out the incomplete data todecision table, discretization and so on. Secondly, in the data reduction stage, it can reduce initialitem-set by attribute reduction algorithm. Thirdly, in the rule generating stage, it can be realizedby attribute value reduction. The three respection shows the importance of rough set theory in thefield of association rule mining fully.Rough sets are applied in the field of association rule mining, which, on the one hand, solvethe key problem of mining useful information; on the other hand, improve the development ofdata mining and speed up the pace of rough set theory. In view of these advantages, this paperincreases association rule mining to a new level from the five aspects of theory, algorithm,experiment, model and instance, which confirms the advantages of rough set theory. Thefollowing task is mainly done by this paper:1. This paper deeply studies related theoretical knowledge of association rules as well asassociation rules mining, and mainly discuss the advantages, existing problems and nextimprovement direction of Apriori algorithm.2. On the basis of summarizing and analyzing the performance and features of Apriorialgorithm, a new data structure is introduced to improve Apriori algorithm according to thedisadvantages of Apriori algorithm and improved algorithm. This paper directly generatefrequent item-set by this data structure, which greatly improves the efficiency of mining itemsets in huge amounts of data. The whole process only scans the database one time, and does notgenerate candidate item sets at all. The running efficiency in the aspect of time is compared byexperiment, which proves the feasibility and effectiveness of new algorithm.3. On the basis of mastering the relationship between association rule mining and rough sets,related knowledge of rough set theory are studied deeply, such as knowledge, knowledge base,Decision information system, knowledge reduction, dependence of knowledge and so on. Thispaper summarize and analyze three different core idea of attribute reduction algorithm as well asthe similarities and differences of idea, and study one of these algorithms hardly. Attributereduction algorithm is optimized by giving attribute importance a new definition of attribute reduction algorithm. New reduction algorithm is applied in instance of association rule mining,which produces good effect by verification.4. A model is constructed, which is called association rule mining model based on rough settheory. And it is added to rough set theory and two kinds of improved algorithm.