Study And Improvement Of Association Rules In Data Mining

Since mankind has entered the information society, with the rapid development of information technology, the amount of data generated by human society is increasing with each passing day. However, how to change complex data into easily accepted knowledge is a problem which is faced by data researchers. Faced with the situation of "Large amount of data, but very few of knowledge", Data mining technology emerged. This discipline which originated in the field of artificial intelligence and database has formed system and played a significant role in practical application.Firstly, this paper expain background and main contents of data mining technology, then, detailed analysis the general flow of data mining and in-depth study the basic idea of classification, clustering, outlier analysis in data mining. In the study of the theory of association rules in data mining, the author first theoretically research Apriori algorithm and FP-tree algorithm, then experimentally verify the advantages and disadvantages of this two algorithms by using weka software. The author innovativly highlight the results of association rules mining based on combination of general data mining processes and theory of association rules by proposing the concept of "mutual confidence". The concept of "mutual confidence" can highlight more meaningful one-dimensional association rules in all results. When negative association rules related theories are researched, author use reverse thinking to innovate a negative association rules algorithm which is based on FP-tree algorithm. Then author describes the idea of the new algorithm and showed the process of the new algorithm according to a practical example and validation experiments. This new algorithm is an useful supplement to the researchment of negative association rules in data mining.