Association Rules Mining Based On Interest Measure

Data Mining is a process which proposed people do not know in advance, but potentially useful information and knowledge from a large number of incomplete, noisy and ambiguous, random data. It is a kind of deep-level data analysis methods, is considered to be one of the effective ways which can solve the "data explosion and knowledge -poor". Mining association rules are an important data mining models, Because it can better capture the important relationship between the data and found that the rules of the form of simple easy to understand, in recent years, it has become a hot field of data mining. This paper analyzed and studied in depth for mining association rules and proposed new rules mining approach. It mainly includes the following:First, in this paper, in-depth study of the Apriori algorithm and it's improved algorithm, in view of the limitations of the algorithm to improve the new algorithm, SHAI-FP-growth algorithm. SHAI-FP-growth algorithm improved the FP-growth algorithm for bad scalability and practicality .The SHAI-FP-growth algorithm joined the division of thought and filtering theory in the FP-growth algorithm. The improved algorithm can effectively reduce the number of candidate sets, and can effectively reduce the number of scanning database, resulting in time complexity and space complexity are improved.Sencond, on the existing degree of interest in an in-depth research, and proposed a new interest-PSA(Piantesky-Shapiro-Accuracy) interestingness in considering the objective and subjective degrees, which combining the merits of objective and subjective Interestingness. In tested the rules based on the new Interestingness are valuable and efficient.Finaly, this paper proposed a association rules mining method based on PSA association rules mining, and analysed the method through a test. The method is based upon the new Interest degree (PSA) and the SHAI-FP-growth algorithm.