Is there any research on the transformation of the clustering results according to the K value?
Basically, the K-medoids algorithm is an algorithm that clusters according to the set value of k. That is, the k value means the number of each medoid, that is, the number of divided sets. When the number of sets is divided, the measured results also change.
In "B. Pardeshi and D. Toshniwal, "Improved k-medoids clustering based on cluster validity index and object density," 2010 IEEE 2nd International Advance Computing Conference (IACC), 2010, pp. 379-384, doi: 10.1109/IADCC.2010.5422924.", which conducted a study on the efficiency of K-medoids, it can be seen that the result value of clustering varies according to the change of k value.
In this paper, the k value was set to 2 to make it easier to classify normal and offensive traffic, and we divided it into normal traffic sets and attack sets, respectively.
