calculation of the distances between the features of the same cluster

Hello,

The k-means syntax

[IDX,C,sumd,D] = kmeans(X,k)

allows you to calculate the

1. within-cluster sums of point-to-centroid distances (sumd)

2. distances from each point to every centroid (D)

How can I calculate the within-cluster sums of point-to-point?

I am referring to the distances between the features (points) that belong to the same cluster, not to the distances between the features and the centroids (sumd).

Also, how can I calculate the distances between the centroids?

Thank you very much.

Best,

Pavlos

Answer 1

[Tom Lane]

Distances between pairs of points in cluster 1:

squareform(pdist(X(IDX==1,:)))

Use something similar to compute distances between pairs of points in C.

Answer 2

[pavlos]

Thank you sir.

Best,

Pavlos