Changing K mean clustering distance metric to canberra distance or any other distance metric on python

问题

How do I change the distance metric of k mean clustering to canberra distance or any other distance metric? From my understanding, sklearn only supports euclidean distance and nltk doesn't seem to support canberra distance but I may be wrong. Thank you!

回答1:

from scipy.spatial import distance
from nltk.cluster.kmeans import KMeansClusterer

obj = KMeansCluster(num_cluster, distance = distance.canberra) 

来源：https://stackoverflow.com/questions/59554641/changing-k-mean-clustering-distance-metric-to-canberra-distance-or-any-other-dis