Cluster evaluation metrics in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding Silhouette Score Interpretation

Which statement best describes what a Silhouette Score close to +1 indicates about a clustering result?

A Clusters have many outliers and points are far from cluster centers

B Clusters are overlapping heavily and points are assigned randomly

C Clusters are well separated and points are appropriately assigned to their clusters

D Clusters are too small and have very few points

💡 Hint

Think about what a high Silhouette Score means for the distance between points and their own cluster versus other clusters.

A Silhouette Score near +1 means points are much closer to their own cluster than to other clusters, indicating good separation and assignment.

❓ Predict Output

intermediate

Output of Adjusted Rand Index Calculation

What is the output of the following Python code snippet using sklearn.metrics.adjusted_rand_score?

ML Python

from sklearn.metrics import adjusted_rand_score
labels_true = [0, 0, 1, 1, 2, 2]
labels_pred = [1, 1, 0, 0, 2, 2]
score = adjusted_rand_score(labels_true, labels_pred)
print(round(score, 2))

A 1.0

B 0.0

C -1.0

D 0.5

💡 Hint

Adjusted Rand Index is symmetric and equals 1 when clusterings are identical up to label permutation.

The predicted labels are a perfect permutation of the true labels, so the Adjusted Rand Index is 1.0.

❓ Model Choice

advanced

Choosing the Best Metric for Clustering with Unknown Labels

You have unlabeled data and want to evaluate your clustering algorithm's quality. Which metric is most appropriate?

A Adjusted Rand Index

B Silhouette Score

C Normalized Mutual Information

D Accuracy

💡 Hint

Consider which metrics require true labels and which do not.

Silhouette Score does not require true labels and measures how well clusters are separated internally.

❓ Metrics

advanced

Interpreting Davies-Bouldin Index Values

Which of the following statements about the Davies-Bouldin Index (DBI) is true?

A Lower DBI values indicate better clustering with compact and well-separated clusters

B Higher DBI values indicate better clustering quality

C DBI values range from 0 to 1, where 1 is perfect clustering

D DBI is only valid for binary clustering problems

💡 Hint

Think about what a low versus high DBI means for cluster similarity and separation.

A lower Davies-Bouldin Index means clusters are compact and far from each other, indicating better clustering.

🔧 Debug

expert

Identifying the Error in Cluster Evaluation Code

What error will the following code raise when executed?

ML Python

from sklearn.metrics import silhouette_score
X = [[1, 2], [2, 3], [10, 10], [11, 11]]
labels = [0, 0, 1]
score = silhouette_score(X, labels)
print(score)

A No error, prints a float score

B TypeError: silhouette_score() missing required positional argument

C IndexError: list index out of range

D ValueError: Number of labels does not match number of samples

💡 Hint

Check if the number of labels matches the number of data points.

The labels list has fewer elements than the data points, causing a ValueError.