engleski

Center based iterative method for clustering and multivariate outlier detection

In this investigation the center based iterative method is constructed by modifying the well known k-means algorithm. The modification is conducted by involving the fact that the Mahalanobis distance follows the chi-squared distribution. The proposed method is presented and examined on the Gaussian mixture model, where a local convergence properties and existence of a fixed point of the proposed method are determined.

clustering, outlier detection, convergence, fixed point

nije evidentirano