Abstract—Ward algorithm is one of the system clustering
methods. The algorithm makes two large classes easily generate
a large distance so as to be not easy to be merged, in contrast, it
makes two small classes generate a small distance and be easy to
be merged. However, the limitation of the method is that it is
difficult for the class that has been obtained to be classified
again. If we first use Ward method clustering samples, then
make each of the obtained classes use PAM algorithm, so that
each class can have a chance to be redivided, so get a more
detailed clustering effect. In view of this idea, the paper proposes
a new clustering method based on Ward and PAM (Ward &
PAM algorithm). The proposed method combines the
advantages of the two algorithms, which makes the clustering
result be more accurate and detailed. Moreover, the paper
optimizes the algorithm index formula. Finally, this paper
makes a detailed comparison analysis of the experimental results.
The experimental result analysis shows that the performance of
Ward & PAM algorithm is better than that of Ward algorithm.
Index Terms—Ward algorithm, PAM algorithm, clustering,
validity index.
Hongmei Nie is with Zhejiang Normal University, Jinhua, China (e-mail:
nhm@zjnu.cn).
[PDF]
Cite: Hongmei Nie, "A New Clustering Algorithm Based on Ward_PAM," International Journal of Future Computer and Communication vol. 5, no. 1, pp. 8-12, 2016.
