Pros | Cons | |
k-means | easy to implement and has high calculation efficiency. | depends on the initial clusters. needs to set the numbers of clusters k initially. |
k-means++ | can decrease the dependency on initial clusters. | needs to set the numbers of clusters k initially. |
x-means | can determine the optimal number of clusters. | depends on the initial clusters. |
x-means++ (Ours) | can determine the optimal number of clusters. can decrease the dependency on initial clusters. | - |