Study On Association Rules Algorithm And Application For Data Mining

Data mining, a hot topic in data base and artificial intelligence fields, has attracted extensive interest in academia and industry. Association rules is one of the key technologies in data mining, which can improve the current status of abundant data but insufficient knowledge effectively. It will be significantly valuable in theoretic studies and application.Usually, the mining of association rules is used for huge database. However, the low efficiency is still its bottleneck in practice. In this paper two approaches are proposed to improve the efficiency of data mining. The first approach is an optimized Apriori algorithm based on the business address index table. This algorithm uses a strategy of reduction affairs worked on the steps of supporting count of Ck. In order to get an efficient processing of reduction affairs, the business address index table is used to make the database orderly. The results show that this method is better than the classical Apriori algorithm, however it still has a limitation that still works for mining of the static data. The second approach is the mining algorithm based on the combined chain of candidate item. This algorithm gives the formulas for the procedure of combination and separation of candidate items. In order to combine the user's requirements and algorithm closely, a linked list is used to create combinations of candidate items and through scanning the business database one time to do mining. The results show the runtime of this algorithm is much better than classical Apriori algorithm. It not only can avoid "missed mining" but also can implement an incremental mining based on restrictive rules.At the end part of this thesis, two specific applications of mining methods for association rules are discussed. First, we realized the association rules mining in a practical application by using the optimized Apriori algorithm based on business address index table, and its purpose was to improve the library room distributing and library collection distributing of Hunan Finance and Economics College. It is of great practical significance to scientifical management of library; secondly, we studied on an application of association rules mining on estimation of the flow of library book transportation by mining algorithm based on the combined chain of candidate item.