A Research On Strong Association Rules And Related Application

In order to make an analysis on shopping basket of supermarket, R. Agrawal, et al. put forward the Association Rules of mining theory and established the framework of Support - credibility in 1993. Over the last decade, this theory has become a very important subject in the field of Data Mining, and attracted great attention of many scholars and experts with practical applications. However, the problem of redundant rules still exists in the theory of Support - credibility, that is, the result set will always take in many boring models and result in many redundant rules.To solve this problem, this paper unprecedentedly sets up the support threshold set by using the penalty function methods for the effective solution to redundant rules. In the genetic algorithm, the paper investigates the basic principles of the algorithm and the solution of TSP, and adopts a new kind of prime factor chromosome coding method and draws in the most frequent distribution form. By changing the original characters into the integer, the coding method changes the string into numerical computing, and reduces the attributes of database to value-based items. The introduction of the most frequent distribution form makes the algorithm be excavated in the most frequent item areas, with the searching space being pruned effectively.Analyzed from the experimental results, the genetic algorithm used in this paper has certain advantages in efficiency and preciseness for the discovery of valuable rules. Meanwhile, this paper also illustrates the application of association rules in mining technology, medical prescription of intelligent supervision, guidance systems and decision-making of case investigation.