Font Size: a A A

Research And Improvement Of Association Rules Algorithm

Posted on:2014-04-02Degree:MasterType:Thesis
Country:ChinaCandidate:Z S DengFull Text:PDF
GTID:2268330428966695Subject:Software engineering
Abstract/Summary:PDF Full Text Request
Data mining has become one of the hot research field in information technology,the tendency is based on the development of computer technology,the appearance of new theory the model and the like。algorithm of the association rules is an important research branch of data mining, many fruitful results about it has been achieved, but the development of mass of data and the emergence of large databases is so fast that to improve the efficiency and accuracy of mining association rules become the urgent need.Tentative research as regards the arithmetic of association rules is studied in the discourse, first the development of data mining technology is introduced briefly, then the concepts of association rules and the problem of the present stage is expounded, the association rules and familiar algorithms are introduced, Apriori algorithm and FP_growth algorithm and Partition algorithm are compared in performance。In the paper, the multi-level association rules and frequent itemsets mining algorithm is analyzed and reviewed according to the research area of association rules,multi-level association rules and frequent itemsets mining rules, Finally, the association rules based on user is analyzed,the metrical method of interest degree is analyzed and compared, and the improved algorithm is put forward.The efficiency of association rule algorithm is described particularly, a series of algorithms that closely rely on it is put forward. the algorithm to solve all problem of mining data is inexistent on account of the complicated diversity of data mining task and the simplification of the algorithm application area, what’s more, many algorithm that have different emphases are exist in the same field. The main aspect in the paper is the efficiency problem and design problem of association rule algorithm.The association rules based on user interest degree is introduced chiefly in this paper, the method of interest measure is analyzed in detail that based on information, probability, correlation. An improved algorithm that its name is Bit-interest is advanced, the definition, format, code of it are described. an example of verification, performance analysis and result analysis are advanced.
Keywords/Search Tags:Data mining, Association rules, The user interest degree, Frequent itemsets mining
PDF Full Text Request
Related items