Reputation:
Using scipy kmeans I can get the centroids of my clusters with
centroids, variance = kmeans(pixel,3)
and also an array showing which value is assigned to which cluster:
code, distance = vq(features, centroids)
But how do I get the actual threshold values separating the clusters from each other? Is there any variable or command containing these?
thanks for any advice
Upvotes: 1
Views: 2135
Reputation: 8715
K-means does not use numerical thresholds.
Every point belongs to the closest cluster, so the "threshold" is the hyperplane (see "Voronoi diagram" in Wikipedia) inbetween of the two nearest cluster centers.
Upvotes: 2