If we use k-means in a dataset where k is equal to the number of points in the dataset, and each cluster is made out of only a point. Considering that we have given a distance method, we can classify a test instance associating it with the cluster whose centroid is nearest to that given point. The centroids are not changing in the test phase.
With what automatic classification method is the above strategy similar to?
Yes, this is a homework. My answer was knn, is this correct? But I do not know how to argue this properly.