Hard Clustering Concepts and Algorithms Answers 1. What effect do outliers have in hard clustering? A. They can lead to more accurate cluster centers. B. They can disproportionately affect the cluster centers. C. They have no effect on the final clusters. D. They simplify the clustering process. Answer: They can disproportionately affect the cluster centers. (B) Outliers can negatively influence the determination of cluster centers in hard clustering. 2. Why is the choice of the number of clusters (k) crucial in hard clustering? A. It controls data preprocessing requirements. B. It determines the algorithm's speed. C. It significantly affects the clustering outcome. D. It influences the initialization of centroids. Answer: It significantly affects the clustering outcome. (C) Choosing the correct number of clusters is essential for effective clustering results. 3. What assumption does hard clustering make about the shape of clusters? A. Clusters do not exist in real-world data. B. Clusters are spherical in shape. C. Clusters can have infinite dimensions. D. Clusters are typically irregular and asymmetric. Answer: Clusters are spherical in shape. (B) Hard clustering algorithms generally assume that clusters are spherical. 4. Which of the following is not a common application of hard clustering? A. Anomaly detection B. Bioinformatics C. Customer segmentation D. Image transformation Answer: Image transformation (D) Image transformation is not a typical application of hard clustering techniques. 5. What is a significant limitation of hard clustering when handling data with non-spherical clusters? A. It tends to underutilize computing resources. B. It decreases the accuracy of the clusters drastically. C. It cannot effectively capture complex shapes in the data. D. It requires advanced machine learning algorithms. Answer: It cannot effectively capture complex shapes in the data. (C) Hard clustering has difficulty in identifying complex or non-spherical clusters properly. 6. What is the defining characteristic of hard clustering? A. Each data point can belong to multiple clusters. B. Data points are categorized into multiple groups based on similarity. C. Clusters are defined based on probabilistic membership. D. Each data point is assigned to exactly one cluster. Answer: Each data point is assigned to exactly one cluster. (D) Hard clustering uniquely assigns each data point to a single cluster, contrasting with soft clustering. 7. Which of the following distance metrics is NOT commonly used in clustering? A. Manhattan distance B. Euclidean distance C. Hamming distance D. Cosine similarity Answer: Hamming distance (C) Hamming distance is typically used for categorical data, whereas the others are common in clustering with numerical data. 8. In k-means clustering, what is the primary goal during each iteration? A. To assign data points based on distance from the nearest centroid. B. To initialize cluster centroids randomly. C. To increase the number of clusters. D. To merge clusters into larger groups. Answer: To assign data points based on distance from the nearest centroid. (A) The main objective during each iteration of k-means is to assign data points to the nearest centroid. 9. Which clustering method uses medoids instead of centroids? A. Hierarchical clustering B. Agglomerative clustering C. K-means clustering D. Partitioning Around Medoids (PAM) Answer: Partitioning Around Medoids (PAM) (D) Partitioning Around Medoids uses actual data points as medoids to represent clusters. 10. What is an important factor in the initialization step of clustering algorithms? A. Assigning data points to identified clusters. B. Determining the maximum number of iterations. C. Setting the initial positions of cluster centers. D. Selecting the distance metric to use. Answer: Setting the initial positions of cluster centers. (C) Initialization is crucial as it determines the starting positions of cluster centers, impacting the algorithm's results. 11. What is a primary advantage of hard clustering? A. It is computationally efficient for large datasets. B. Produces results that are often computationally intensive. C. Allows for partial membership in multiple clusters. D. Clusters are not easy to interpret. Answer: It is computationally efficient for large datasets. (A) Hard clustering is computationally efficient, making it suitable for large datasets while being straightforward. 12. How does agglomerative hierarchical clustering start? A. By merging the closest two clusters. B. With each data point as an individual cluster. C. With a single cluster encompassing all data points. D. By assigning random initial cluster centers. Answer: With each data point as an individual cluster. (B) Agglomerative clustering begins with each data point treated as a separate cluster, progressively merging them. 13. What is the main purpose of recalculating centroids in k-means clustering? A. To determine the final number of clusters. B. To assess the initial cluster assignments. C. To decrease the distance metric for all points. D. To refine cluster assignments based on new data point groupings. Answer: To refine cluster assignments based on new data point groupings. (D) Recalculating centroids helps refine cluster assignments by ensuring they represent the mean of newly assigned data points.