breast-cancer-wisconsin.data or you can go to find it on GitHub here
The data set contains 11 columns, separated by comma. The first column is the example id, and has been ignored. The second to tenth columns are the 9 features, based on which K-means algorithm works. The last column is the class label, and has been ignored as well.
K-Means algorithm performs clustering on the above dataset with K = 2, 3, 4, 5, 6, 7, 8. For each K value, the algorithm is first run and then the potential function is computed as follows:
where m is the number of examples, xj denotes the feature vector for j th example and µC(j) refers to the centroid of the cluster that xj belongs to.
Empty clusters in a certain iteration have been droped and randomly the largest cluster is split into two clusters to maintain the total number of clusters at K.
A graph is plot for the values of k and L(K).