Research On Data Mining Techology In Performance Management

Nowadays, information techology is developed. If a company want to survive in the society in which information expand, it need implement Enterprise Resource Planning (ERP). ERP will be the important means by which the companies implement the management information stratagem. Decision support basing on company data becomes the developing trend and ultimate target. Data mining is the important tool realizing the goal. To Performance Management that is important part of ERP, the intelligentized need is more exigent. So research on data mining techology in performance management is more valuable.Frequent itemset mining plays a crucial role in Association Rules Mining. Because the size of the set of all the maximal frequent itemsets is much smaller than that of the frequent itemsets, it has been observed that it suffices to mine only the set of maximal frequent itemsets instead of every frequent itemsets. The improvement in the field of maximal frequent itemsets mining will affect that in the field of Association Rules Mining.Then a algorithm called FP-MaxgrowthforKPI is proposed for mining maximal frequent patterns in the Key Performance Index (KPI) confirming in performance management, through analyzing many maximal frequent itemsets mining algorithms and integrating the fact of performance management.The algorithm adopts bitmap data format; according to users'requirements and the data characteristics in performance management, we can get the FP-tree by using this algorithm. This algorithm has improved traditional algorithms to accelerate the generation of maximal frequent itemsets in order to adapting to the application in performance management, through analyzing the merits and disadvantages of Apriori and FP-growth methods, integrating the method with several prune techniques and local maximal frequent itemsets for superset checking quickly. The experiment results show the effective performance, which are better than the traditional algorithm.A model based on maximal frequent itemsets mining is presented that specify the Key Performance Index (KPI) in performance management. Finally, an example has been given to show the feasibility and efficiency.