Hierarchical clustering (linkage) in SciPy - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is hierarchical clustering?

Hierarchical clustering is a method to group similar data points into clusters by building a tree of clusters. It does not require a fixed number of clusters beforehand.

beginner

What does the 'linkage' function in scipy do?

The 'linkage' function in scipy calculates distances between clusters and merges them step-by-step to form a hierarchy, producing a linkage matrix that shows how clusters combine.

intermediate

Name three common linkage methods used in hierarchical clustering.

Three common linkage methods are: 1) Single linkage - distance between closest points, 2) Complete linkage - distance between farthest points, 3) Average linkage - average distance between all points in clusters.

beginner

What is the output of the scipy linkage function?

The output is a linkage matrix, a 2D array where each row shows which clusters were merged, the distance between them, and the number of original points in the new cluster.

intermediate

Why is hierarchical clustering useful compared to k-means?

Hierarchical clustering does not need you to choose the number of clusters in advance and shows the full cluster tree, which helps understand data structure better.

What does the 'linkage' function in scipy.cluster.hierarchy return?

A A linkage matrix showing cluster merges

B A list of cluster labels for each data point

C A distance matrix between all points

D A dendrogram plot

The linkage function returns a linkage matrix that records how clusters are merged step-by-step.

Which linkage method uses the shortest distance between points in clusters?

A Complete linkage

B Ward linkage

C Single linkage

D Average linkage

Single linkage uses the shortest distance between points in two clusters.

In hierarchical clustering, what does the dendrogram represent?

A A plot showing cluster centers

B A tree showing cluster merges and distances

C A heatmap of data values

D A scatter plot of data points

A dendrogram is a tree diagram that shows how clusters merge at different distances.

Which of these is NOT a valid linkage method in scipy?

A Single

B Complete

C Median

D K-means

K-means is a different clustering algorithm, not a linkage method.

What is the main advantage of hierarchical clustering over k-means?

A It does not require specifying the number of clusters

B It is faster for large datasets

C It always produces spherical clusters

D It uses random initialization

Hierarchical clustering builds a cluster tree without needing the number of clusters upfront.

Explain how the linkage matrix represents cluster merges in hierarchical clustering.

Think about what information you need to know how clusters combine step-by-step.

Describe the difference between single, complete, and average linkage methods.

Focus on how distance between clusters is measured.