Kmeans ++ algorithm
Kmeans ++ algorithm, can solve the main problem of selecting the initial centers, can not solve the number k.
Kmeans ++ main idea is to select the initial cluster centers to try far.
practice:
1. The input data points randomly selected as a first cluster center.
2. For all of the data points, calculate its nearest cluster center distance D (x)
3. re-select a data point as the center of the cluster, selection principles: (x) the point D is larger the larger the probability of the chosen cluster center.
4. Repeat 2-3 until the selected k cluster centers.
5. Run Kmeans algorithm.