The Research And Applications Of Data Mining Algorithms Oriented Incremental Updating

Due to the problem that the traditional data mining algorithms orient to static data and the data in the database updates day by day, If the result of data mining could not updated on time, it will influence the strategy or right judgement. So it has prodigious significance to research on the data mining algorithms for the incremental updating. This paper mainly aims at the research on association rules and clustering algorithms for the incremental updating.A new incremental updating association rules algorithm PFUP is presented in the paper. It joins the strong large itemsets into the small quantitative of candidate itemsets on the basis of strong large itemsets concept, and adopts the early pruning strategy to cut down the times of scan of database. The simulation shows that the execution time of PFUP algorithm has reduced to about 50% compared with FUP algorithm in the case of the same database and support. Then, on the basis of the PFUP algorithm, the measure of reducing transactions is adopted. The Simulation shows that the new algorithm has the better superiority.This paper focuses the research on the incremental clustering mining algorithms based on density. Firstly, it introduces the basic concept and the idea of DBSCAN algorithm, and shows the recent two algorithms of the incremental clustering algorithm based on density. Because the incremental clustering algorithm just considers the situation of increasing data or deleting data, a new incremental clustering algorithm UDBSCAN is put forward, in which the MinPts changes. The new algorithm proves the superiority by the Simulation.Finally, an OLAM application model has been designed and implemented in the system of Medical Insurance .The model employs the medical data of 2005 from medical insurance data base of some city in zhejiang province and builds data warehouse on the theme of medical insurance and diagnosis. It analyzes the data from multidimensional viewpoints and uses the incremental association rules algorithms to gain the valuable information timely.