Grid Probability Density-based clustering for uncertain data streams over sliding windows

The existing algorithms for clustering uncertain data streams are unable to obtain clusters of arbitrary shapes. In order to address this issue, this paper proposes GD-CUStreams, which adopts two phased-based clustering frameworks. In the online phase, the store space is divided into grid and Uncertainty Grid Clustering Feature (UGCF) is defined to acquire the uncertainty information of tuple and the summary information is stored in UGCF. In the offline phase, according to the Grid Probability Density threshold, GD-CUStreams detects all grids. Furthermore, the type of grid is determined and sporadic grids will be detected from all sparse grids based on density threshold function. While the clustering request arrives, GD-CUStreams outputs all grids with the type of normal and transition. Finally, clusters of arbitrary shapes are generated. Experimental results show that GD-CUStreams has higher clustering quality. ICIC International