Research On Data Stream Clustering Algorithm Based On Density Grid Over Sliding Window

This paper proposes a sliding window density grid-based data stream clustering algorithm SD-Stream. Proposed "scribe degree" concept, by calculate the "scribe degree" of data element,Redefine the grid of data element should be belongs to. Improved the idea of density grid algorithm which is simply based on the number of data points within the grid as the absolute density, solved two data points belong to two clusters but in a grid and the edge of grid-based algorithms lead to information loss problems, which not only use the high efficiency of density grid based algorithm, also improved the clustering accuracy. Use a sliding window to storage summary information, solved the time information loss problem of density grid-based data stream clustering algorithm which the efficiency independent of its data elements within the sliding window, can respond to user different time granularity clustering requests. The experimental results show that, SD-Stream algorithm can find clusters of arbitrary shape, compared with similar algorithms the implementation efficiency and clustering accuracy are improved, and have good adaptability and feasibility.