One of the issues with the K-means algorithm is that if we choose initial centroids randomly, the clustering result may not be good. We discuss in the class
that we can avoid this by running the K-means multiple times and choose the result with the highest quality based on some criterion, such as SSE. Another
way of resolving this issue is to select initial centroids in a non-random way. Discuss different ways of choosing initial centroids that would increase the
quality of clustering result.
Sample Solution
