Data Mining Of Association Rules Algorithm

Rapid development of modern technology has brought human beings a lot of convenience as well as large amount of data. To better understand these data and make it available for people to use, KDD and DM technology emerged. Mining Association Rules of DM is an important research direction and the mining of is the core problem of Mining Association Rules. In this paper, I discussed the association rules and the classic Apriori algorithms in detail, and summed up the characteristics and limitations of classic mining algorithms. On this basis I raised two improved algorithm -- Association Rule Mining algorithm based on Granular Computing and Association Rule mining Algorithm based on Genetic Algorithms.The Association Rule Mining algorithm based on Granular Computer, which imports granular computing thought, takes each item as information granular and obtains the bitmap expression of items through scanning databases just one time. It calculates Support of item set using "AND" and "OR" operation and insert the qualified items into the frequent K concentration using Cycle judgment method without generating candidate items C_k. The advantage of this algorithm is that it just needs scanning database once and just storing frequent set which improves the efficiency of association rule mining by reducing time-consuming and memory-taking. Meanwhile, focusing on the problem that it is difficult to find strong association rules in the lower or primitive layer between data items for many applications, this paper raised a multidimensional multilevel association rule mining algorithms based on the granular computing. In the concept of multidimensional attributes Stratification, this Algorithm mines association rule using Rule Mining algorithm based on Granular Computing on each level. Moreover, this algorithm applies level relation of tablets when calculates Support of the item set on high level and accordingly, improves the efficiency of association rule mining.Mining association rules algorithm based on genetic algorithm is the algorithm which improves the traditional genetic algorithms and is applied in association rule mining. This algorithm uses an adaptive mutation rate method which makes the high fitness individual in the early evolution stage avoid falling into local optimal solution because of excessive reproduction. Furthermore, this algorithm improves the individual selection and is quite practical. Finally, this algorithm was applied into Mining Association Rules relevant experiments which verified the high efficiency and reliability of Mining Association Rules based on genetic algorithm when dealing with large-scale projects sets.