Improved Weighting K-Means Algorithm Based on Covariance Matrix

Feature weighted clustering algorithms are an important direction of research in cluster analysis. However, the existing weighted k-means algorithms have some problems that the influence of noise features cannot be excluded and there exist some defects in the use of feature weights. To solve these problems, this paper firstly uses covariance matrix to weight the features of data sets; then this paper presents a theoretical analysis and a proof; lastly, the clustering results of the proposed algorithm were compared and analyzed with existing algorithms on a real datasets, which shows that the proposed algorithm can effectively eliminate the influence of noise features and improve the clustering performance. Through this research, the proposed algorithm can effectively improve the clustering performance and has a wide range of application prospects.