I have been researching and studying about partition-based clustering algorithms like K-means and K-Medoids. I have learned that K-medoids is more robust to outliers compared to
To prevent bad things from happening (infinite loops etc.) always prefer the cluster the point already is assigned to when tied.