Clustering on Uncertain Data Stream over Sliding Windows

This paper proposes a density grid-based algorithm (C_UStream) for clustering on uncertain data stream in sliding window which can find clusters of arbitrary shapes. The statistical summary information of each grid is stored in linked queue structure by using sampling window mechanism. In order to guarantee the validity of clustering, the expired grids in the current window are removed regularly. Furthermore, a dynamic sporadic grids deletion mechanism is developed to delete most of outliers periodically which greatly improve the space and time efficiency. The experimental results on the synthetic and real data sets show that C_UStream has superior clustering quality and efficiency than other similar algorithms.