ABSTRACT
Step 1: Randomly initialize the k cluster center. Calculate the cluster center list C = [c1, …, ck]. Step 2: Partition each data object into the nearest cluster Ci, yields:
If dj ∈ Ci, subject to for any m, where m = 1, …, k and m ≠ i and
(2)
Step 3: Recalculate the cluster center list C based on the current partition,
c N
jd Cj i = ∑1 (3)
Repeat Step 2 and Step 3 until there is no change for each cluster or the sum-of-squared-error less than the threshold.
