MLS-C01 Question 117
Select 3You are working on a project to segment customer data into distinct groups based on purchasing behavior. After performing K-Means clustering in Amazon SageMaker, you want to determine the optimal number of clusters. Which of the following approaches should you use to achieve this?
- A
Use the elbow plot to identify the point where the within-cluster sum of squares (WCSS) starts to level off.
- B
Perform hierarchical clustering to visualize the dendrogram and identify cluster hierarchies.
- C
Measure the silhouette score for different values of K and select the K with the highest silhouette score.
- D
Analyze the cluster centroids to manually decide the number of clusters.
- E
Use the Davies-Bouldin Index to evaluate clustering performance for different values of K.
Show answer and explanation
Correct answers: A, C, E
Explanation
Determining the optimal number of clusters is a key step in clustering analysis. The elbow plot, silhouette score, and Davies-Bouldin Index are all widely used methods for evaluating clustering performance and identifying the ideal number of clusters. Hierarchical clustering and manual analysis of centroids, while valuable for other purposes, are not standard approaches for determining cluster count in K-Means.
- A. Correct.
Using the elbow plot is a common method to identify the optimal number of clusters by finding the point where the WCSS curve starts to flatten, indicating diminishing returns in reducing variance by adding more clusters.
- B. Incorrect.
Hierarchical clustering and dendrograms are useful for visualizing cluster hierarchies but are not directly used to determine the optimal number of clusters for K-Means.
- C. Correct.
The silhouette score measures how similar an object is to its own cluster compared to other clusters and is effective for finding the optimal number of clusters.
- D. Incorrect.
Analyzing the cluster centroids manually can provide insights but is not a systematic or reliable approach to determining the optimal number of clusters.
- E. Correct.
The Davies-Bouldin Index is a metric for evaluating clustering performance, and minimizing this index can help identify the optimal number of clusters.