An improved rough k-means clustering algorithm

Rough k-means clustering algorithm proposed by Lingras is sensitive to the initial centers of the k cluster and outliers and may result in identical clustering and non-convergence. In this paper, an improved rough k-means clustering algorithm is proposed. The k objects with maximum potentials are chosen as initial centers. The absolute distance between object and center of clusters is considered to decide whether a data object belongs to the lower or upper approximation set of a cluster, so the division of boundary area is more reasonable. General classification accuracy considering the objects in lower approximation set and boundary area is defined for rough k-means clustering algorithm, and it is more appropriate for evaluating rough k means clustering. The simulation results show that, the proposed algorithm has the advantages of high classification accuracy and fast convergence, and can also avoid the bad influence of outlier.