A novel sample weighting K-means clustering algorithm based on angles information
Lei Gu · 2016
One identical weighting scheme for each sample of one cluster is often employed in the traditional sample weighting k-means clustering. However, this paper proposes a novel sample weighting k-means clustering algorithm based on angles information(SWKMA). In this presented SWKMA, firstly, samples of one cluster is divided into two types according to angles information, and secondly, different weighting schemes are used for different types of samples respectively. To evaluate the effectiveness of SWKMA, experiments are done on 11 artificial and real datasets. Experimental results demonstrate that SWKMA can acquire the better clustering performance than k-means, fuzzy c-means and one sample weighting k-means clustering method.