Let's say that we have 4 data points with x1 = 1, x2 = 3, x3 = 6, and x4 = 7. We use a K-means clustering algorithm to group them into 3 clusters. This means the clusters should be x1 and x2 in one, x3 in another, and x4 in the third. How can we by changing the center initialization get a suboptimal cluster assignment that cannot be further improved? Please show/explain in detail please. Thanks.