Research Of Incremental Data Mining Based On Clustering

Clustering analysis is always main aspect of DM(Data Mining) research, and the research of several DM arithmetic based on clustering is also in focus. But at present most of the clustering arithmetic styles are applied for static data sets, in other hand, as for the dynamic data sets, the only way is to redo the process of clustering on the whole data set, so as the quantity of data set becomes more and more huge, and demanding for DM in time, the incremental DM become more and more attractive.The incremental DM based on clustering inherits the former clustering result, and then check the new added data item one by one, in this way, a great amount of computation can be avoided. So the computer system resources can be saved too. Especially when the data set is unexpectedly huge, the incremental DM can show its excellence, such as the instant and valuable information rooted in DM process for users.Firstly, This paper sums up the main research achievement of DM and clustering analysis, and gives the detailed arithmetic theory of DBSCAN which is based on fuzzy. And then gives the correspond incremental clustering arithmetic: Incremental DBSCAN. On account of the processing manner of check the data item one by one, it introduce the batch arithmetic in order to mostly improve the efficiency.At last, the DBSCAN and Incremental DBSCAN arithmetic are approved accordant with each other, besides, the Incremental DBSCAN is much more efficiency with great data set.