Correlative Research On Association Rules Data Mining Algorithm

In recent years, many people in information industry pay more attention to the data mining technique, which is the necessary result of the conflicting movement between the rapid-increasing data and the lack of information day by day. Researching the data mining technique systematically, deeply and detailedly is an objective requirement of the development of the global information. One important area of many researching areas in data mining is the association rule, which has great value in application. This thesis is focus on the correlative study of association rules data mining algorithm.The evaluation criteria are based on support and confidence in existing association rule mining algorithms. But, in many time, the mined association rules with high support and confidence are useless. At the same time, the evaluation criterion does not consider whether the corresponding rules with negative item are useful or not, when the positive item rules with high support and confidence are useless. The thesis introduces the interesting threshold into association rules, which will be used to prune the useless association rules together with the threshold of support and confidence. And based on the predecessors' improved definition of association rules and correlative interesting, the author improves the association rales mining algorithm based on support, confidence and interesting.With the development and improvement of the database management system, more and more inaccurate data are stored in database, which takes great challenge to the applications of data mining technique. The traditional data mining algorithms, such as Apriori algorithm and its improved algorithms, are focus on the mining in accurate concepts, and can not mine the inaccurate or fuzzy concepts. So, the thesis integrates the fuzzy-set concepts with data mining algorithms to make a deeper research on association rules mining algorithms, and introduces the concept of fuzzy association rules, which expresses the associated relationship by fuzzy concepts and extends the range of the application and description of association rules. The author proposes a fuzzy multilevel association rules mining aigorithm based on the multilevel association rules mining algorithm, which will be used to mine association rules in fuzzy concepts.