A hybrid clustering algorithm based on grid density and rough sets

According to the characters of dynamic and SOM clustering algorithm, propose a novel clustering method, rough dynamic clustering based on grid-density algorithm (GDRDC). The algorithm contains initial clustering stages and precise adjustment stages. During switch from the first stage to second stage, according to rough sets idea, class kernel and freedom point sets base on grid-density are determined, and though which the two stages are joined. Then making farther adjustment by dynamic clustering method, the final clustering result is get. The experiment result shows that it is better than SOM and K-means, especially for nonlinear separable data.