International Conference on Information Technology and Computer Science, 3rd (ITCS 2011)
118 The Clustering Algorithm of Evolutional Data Stream Based on Density
Download citation file:
- Ris (Zotero)
- Reference Manager
The clustering algorithm and its analysis are very important methods in data mining. Under the sliding window model of the data stream, this paper presents a new density-based clustering algorithm DSJstream, introduced the concept of density, so it can find any shape, any number of clusters, it can analyze and compare the history of the evolution of data stream. At the same time, by the fading strategy and the granularity adjustment strategy, it is to clear up the forgetting problem of historical data stream. It is very good to deal with noise, save memory effectively and prevent memory overflow. Through experimental analysis proved that the algorithm has good scalability, effectiveness. It is suit for the clustering and analysis under the fast flow rate, large amount of data environment.