Why Clustering using PAM method often attributes one point in the middle of one cluster to another cluster? (for example here K-means clustering )
I have the similar results with my set of data.
Why Clustering using PAM method often attributes one point in the middle of one cluster to another cluster? (for example here K-means clustering )
I have the similar results with my set of data.
I do not know how ClusteringComponents's PAM is implemented, but I strongly suspect that what you observe is an effect of the stopping criteria of the PAM algorithm. E.g. using cluster samples to compute the Silhouette measure might produce these kind of results.