Hierarchical Clustering Quiz

Answers and explanations appear as you take the quiz.

1. What sequence does agglomerative hierarchical clustering build?
   • Repeated cluster merges from individual observations toward larger groups
   • Repeatedly split one cluster into smaller groups from the top down
   • Assign each point to a fixed centroid and update those centroids
   • Join each point to its nearest neighbor once without building nested groups
2. Which rule defines single linkage between two clusters?
   • The maximum pairwise distance across clusters
   • The minimum pairwise distance across clusters
   • The average of all pairwise distances
   • The increase in within-cluster sum of squares
3. Which linkage uses the farthest cross-cluster pair?
   • Average linkage
   • Single linkage
   • Complete linkage
   • Ward linkage
4. What does a dendrogram merge height encode?
   • The p-value for a test that the two groups differ
   • The number of observations contained in the merged branch
   • The proportion of observations assigned to the left child
   • The linkage distance or criterion value at that merge
5. Why can a large jump in dendrogram height be useful?
   • It proves the exact number of natural populations
   • It may suggest candidate cuts that keep separated branches apart
   • It means every leaf is an outlier
   • It identifies the optimal feature scaling
6. Why may single linkage create a chaining pattern?
   • It requires all points to be equally distant
   • It uses the average distance between every pair
   • A sequence of near neighbors can connect otherwise elongated groups
   • It minimizes centroid variance
7. What does Ward linkage seek to minimize when merging?
   • The entropy of labels
   • The largest single pairwise distance only
   • The number of leaves in the tree
   • The increase in within-cluster sum of squares
8. Why standardize features before Euclidean hierarchical clustering when units vary greatly?
   • Otherwise a large-scale feature can dominate distances
   • Scaling changes the dendrogram into a supervised model
   • It forces average linkage to equal Ward linkage
   • It guarantees that the chosen cut is stable