Lecture 12 Evaluating Clustering Flashcards
1
Q
Supervised evaluation
A
Unclear assumption
Impossible to do in practice
Compare partitions, not labels
2
Q
Contingency matrix
A
3
Q
Randindez
A
Computed of Partitions c1 c2 over the same element
4
Q
Adjusted rand index( ARI)
A
(-1,1)
5
Q
Mutual information
A
Not normalized or adjusted
6
Q
Unsupervised evaluation
A
Silhouette score
A= mean distance to sample in same cluster b=….. in nearest cluster
