Find the best scheme using k-means

Question

I have following random data, and I have 2 schemes 1 and 2

1: contains clusters C1 with records {1,2,3} and C2 wit