Data Streams Clustering Algorithm Based on Grid and Particle Swarm Optimization

The offline components of CluStream clustering algorithm based on distance, and it is difficult to find non-spherical character of the cluster. This paper proposes a data streams clustering algorithm based on grid and particle swarm optimization, the algorithm based on two-tier structure of CluStream clustering algorithm. The grid feature vector to represent a snapshot, the grid density, and grid merging technologies is applied in this algorithm. So we can found any non-spherical clusters,In this algorithm.Non-dense grid is periodic and dynamic way to remove,which is good for reducing the space complexity. Using the PSO optimized clustering results in the offline components, in order to get a more precise clustering efficiency. Experiments show that this algorithm is more efficient than the CluStream algorithms, it has a good number of dimensions scalability,and it can find non-spherical nature of the clustering results.