.
Suppose that a set of OTUs measured on n characters have
been clustered by K-means into three clusters A, B, and C, with centroids
defined by {ai}, {bi}, and {ci}, respectively. Values for within-cluster sums of
squares for A, B, and C are s2
A, s2
B, and s2
C, respectively.
a. Write the statistic(s) that should be calculated for classifying OTU X into
one of the three existing clusters.
b. What statistic(s) could be used to determine whether X should be placed
separately into a cluster different from A, B, or C?