Research on Clustering Method Based on Weighted Distance Density and K-Means
Wei Yang, Hua Long, Lihua Ma, Huifang Sun
Abstract
In this paper, the effect of the initial clustering center selection on the performance of the K-means algorithm is studied, and the performance of the algorithm is enhanced through better initialization techniques. In the K-means clustering process, when calculating the density of a data set by using a weighted distance density calculation method, significant improvement in the defects of poor clustering results caused by the local optimum and large intra-cluster variance in the traditional K-means clustering algorithm has been found. Experimental results show that by using the improved method proposed in this paper, the intra-cluster variance of clustering results is reduced by 15.5% compared with the traditional method, which makes great improvement in the performance of the algorithm.