Research on Optimization of Clustering by Fast Search and Find of Density Peaks.

CFSFDP is a clustering algorithm based on density peaks, which can cluster non-spherical data sets, and also has the advantages of fast clustering and simple realization. However, the global density threshold dc, which leads to the decrease of clustering quality, is specified without the consideration of spatial distribution of the data. Moreover, the data sets with multi-density peaks cannot be clustered accurately. To resolve the above shortcomings, an optimization of CFSFDP algorithm based on grid (GbCFSFDP) is proposed. To avoid the using of global dc, first, the algorithm divides the data sets into smaller partitions by using the method of grid partitioning and performs local clustering on them. Then, GbCFSFDP merges the subclasses. Finally, data sets which are unevenly distributed and exist multi-density peaks are correctly classified. Simulation experiments of two typical data sets show that the proposed GbCFSFDP algorithm is more accurate than CFSFDP.