Clustering Methods Quiz

What is the primary aim of clustering models?

• To visualize data in a two-dimensional space

• To classify records into distinct categories

• To subdivide records into homogeneous groups

• To analyze the relationships between variables

Which of the following is a requirement for clustering methods?

• They must always produce spherical clusters

• They must require a predetermined number of clusters

• They should only work with numerical data

• They should be robust to small changes in data

What do partition methods in clustering do?

• They create a tree structure of clusters

• They subdivide a dataset into a predetermined number of subsets

• They identify clusters based on density

• They analyze the grid structure of observations

Which clustering method does not require the number of clusters to be predetermined?

• K-medoids

• Hierarchical methods

• K-means

• Partition methods

What is a characteristic of density-based clustering methods?

• They derive clusters from the number of observations in a neighborhood

• They require a fixed number of clusters

• They are not robust to outliers

• They can only identify spherical clusters

What is the silhouette coefficient used for in clustering?

• To determine the number of clusters

• To evaluate the quality of clusters

• To measure the distance between clusters

• To visualize the clusters in a graph

Which of the following is true about grid methods?

• They do not require any preprocessing of data

• They can only be applied to numerical data

• They achieve reduced computing times but lower accuracy

• They require a large number of attributes

What is the main disadvantage of using samples of reduced size in clustering?

• It can only be applied to small datasets

• It requires more time to compute clusters

• It may lead to lower robustness of the clusters

• It increases the computational complexity

What does the term 'fuzzy methods' in clustering refer to?

• Methods that require a fixed number of clusters

• Methods that are only applicable to numerical data

• Methods that only allow complete clustering

• Methods that assign observations to multiple clusters with weights

Which distance measure is commonly used for numerical attributes?

• Jaccard coefficient

• Euclidean distance

• Cosine similarity

• Hamming distance