An Improved PAM Clustering Algorithm Based on Initial Clustering Centers

Abstract

This paper proposes a novel clustering algorithm with the steady initial center based on the PAM algorithm. As the PAM algorithm arbitrarily chooses the initial centers, this new algorithm improves the location of the initial center, which makes the initial choice much closer to the data distribution. It also overcomes the blindness of generating initial central points. The experimental results show that this algorithm is effective which performs better in classification efficiency and quality, and it also decreases the number of iterations.