Big Data

K-means Clustering Algorithm Characteristics Differences Based on Distance Measurement

Date Added: Dec 2012
Format: PDF

A distance measure for similarity estimation based on the differences is presented through the authors' proposed algorithm. This kind of distance measurement is implemented in the K-means clustering algorithm. In this paper, a new Minkowski distance based K-means algorithm called Enhanced K-Means Clustering algorithm (EKMCA) is proposed and also demonstrates the effectiveness of the distance measurement, the performance of this kind of distance and the Euclidian and Minkowski distances were compared by clustering KDD'99 Cup dataset. Experiment results show that the new distance measure can provide a more accurate feature model than the classical Euclidean and Manhattan distances.