Clustering Algorithm Of Data Stream Base On Fast Search And Find Of Density Peaks

Data stream clustering is a clustering analysis for unbounded sequences of data objects that are continuously generated at rapid rates. We need to satisfy with some characteristics and restrictions of the data stream when the clustering algorithms are applied to the data stream. In the current, the traditional clus-tering algorithms like DBSCAN algorithm, KMEANS algorithm and hierarchical clustering algorithm can be used to data stream clustering. In this paper, we use a new clustering method from" Clustering by fast search and find of density peaks" which is proposed by Alex Rodriguez and Alessandro Laio in 2014. We select the parameter dc automatically and define the density of data points as Gaussian kernel density. Moreover, the noisy of data streams were identified automatically. And then, we extend the algorithm to the data stream by combining with the sliding window model. Compared with classical data stream clustering algorithm-s, the algorithm in this paper improves the accuracy of clustering greatly, and reduces the number of parameters and obtains the recognition of noisy. In this paper, we through simulation and a real data to study our algorithm.