Introduction to Hierarchical Clustering Answers 1. What does the height of fusion in a dendrogram represent? A. The computational cost of clustering B. The average size of clusters C. The dissimilarity between merged clusters D. The total number of clusters Answer: The dissimilarity between merged clusters (C) The height of fusion indicates how dissimilar the clusters are before they merge. 2. Which factor is NOT typically considered when choosing a hierarchical clustering method? A. Desired structure of the clusters B. Data characteristics such as high dimensionality C. Type of visualization needed for results D. Computational resources available Answer: Type of visualization needed for results (C) While visualization is important, it is not a primary factor in selecting a hierarchical clustering method. 3. Which of the following is an application of hierarchical clustering? A. Predicting future trends in stock prices B. Image compression for faster download C. Calculating average distances between data points D. Grouping customers based on purchasing behavior Answer: Grouping customers based on purchasing behavior (D) Hierarchical clustering is commonly used for customer segmentation based on behavior. 4. What is one significant disadvantage of hierarchical clustering? A. It can visualize clusters effectively through dendrograms B. It guarantees optimal cluster formation in any scenario C. It is computationally intensive for large datasets D. It is able to uncover clusters of varying shapes Answer: It is computationally intensive for large datasets (C) Hierarchical clustering can be computationally expensive, especially with large datasets. 5. In terms of cluster structure, which consideration is crucial when choosing a hierarchical clustering method? A. Number of variables in the dataset B. Presence of outliers in the dataset C. Desired uniform size and shape of clusters D. Distribution of data across clusters Answer: Desired uniform size and shape of clusters (C) The desired structure of the clusters, including size and shape, can significantly influence method selection. 6. What is the primary goal of hierarchical clustering? A. To evaluate the performance of supervised learning models. B. To build a hierarchy that best reflects the inherent similarity of data points. C. To randomly group data points without any structure. D. To classify data points into predefined categories. Answer: To build a hierarchy that best reflects the inherent similarity of data points. (B) Hierarchical clustering aims to create a hierarchy that accurately represents the similarities among data points. 7. Which method in hierarchical clustering starts with each data point as a separate cluster? A. Single Linkage B. Agglomerative Clustering C. Average Linkage D. Divisive Clustering Answer: Agglomerative Clustering (B) Agglomerative clustering begins with each data point as an individual cluster and merges them iteratively. 8. What does complete linkage in agglomerative clustering do? A. Merges clusters while considering the average distance between points. B. Merges clusters based on the furthest distance between points. C. Merges clusters based on the nearest neighbor distance. D. Generates clusters without regard to distance metrics. Answer: Merges clusters based on the furthest distance between points. (B) Complete linkage merges clusters based on the furthest distance between points in the clusters. 9. What is a characteristic of divisive clustering? A. It seeks to randomly distribute data points into multiple clusters. B. It continuously merges clusters until one remains. C. It starts with all data points in one cluster and splits them recursively. D. It utilizes only Euclidean distance for all calculations. Answer: It starts with all data points in one cluster and splits them recursively. (C) Divisive clustering begins with a single cluster and recursively splits it into smaller sub-clusters. 10. How does agglomerative clustering determine which clusters to merge? A. By selecting the largest clusters available. B. Through a predefined set of rules without regard for data similarity. C. By using a random selection of clusters. D. Based on the distance metric and linkage criteria. Answer: Based on the distance metric and linkage criteria. (D) Agglomerative clustering merges clusters according to the chosen distance metric and the specified linkage criteria. 11. What role do similarity metrics play in agglomerative clustering? A. They select the initial clusters before any merging occurs. B. They ensure that clusters are equal in size. C. They measure the distance between clusters and influence merging decisions. D. They determine the order in which clusters are displayed visually. Answer: They measure the distance between clusters and influence merging decisions. (C) Similarity metrics are crucial as they measure distances and guide the decisions on cluster merging. 12. Which of the following statements about average linkage is correct? A. It is the fastest method for clustering. B. It does not consider distance in its clustering process. C. It finds an average resemblance between all pairs of data points in the clusters. D. It merges clusters based on the furthest points between them. Answer: It finds an average resemblance between all pairs of data points in the clusters. (C) Average linkage merges clusters based on the average distance of all data points between the two clusters. 13. What is a potential drawback of using single linkage in agglomerative clustering? A. It can lead to the formation of chained clusters that misrepresent the data structure. B. It is computationally less intensive than complete linkage. C. It requires a large amount of data for accurate results. D. It always produces the most accurate cluster representations. Answer: It can lead to the formation of chained clusters that misrepresent the data structure. (A) Single linkage can be sensitive to outliers, leading to chains of clusters that may not represent the overall data structure well.