Which clustering technique partitions data by assigning points to the nearest centroid and requires specifying the number of clusters?

Which clustering technique partitions data by assigning points to the nearest centroid and requires specifying the number of clusters?

• A. DBSCAN
• B. Hierarchical clustering
• C. K-means clustering
• D. Gaussian Mixture Models

Answer: (C)

Partitioning data into a fixed number of groups by assigning each point to the nearest center is the hallmark of K-means clustering. In this method you decide how many clusters you want (the value of k) up front. The algorithm then alternates two steps: assign every point to the closest centroid, creating hard cluster memberships; and recompute each centroid as the mean of all points assigned to it. This process repeats until assignments stop changing, effectively minimizing the within-cluster variance for the chosen number of clusters. This approach stands apart from other clustering methods. DBSCAN groups points based on density and can discover clusters of arbitrary shape without a predefined number of clusters; it relies on density parameters and can label points that don’t belong to any dense region as noise. Hierarchical clustering builds a tree of clusters and lets you choose how many clusters to keep by cutting the tree at some level, rather than fixing k at the start. Gaussian Mixture Models use probabilistic, soft assignments to components and model the data as mixtures of Gaussians, with cluster membership determined by posterior probabilities rather than the nearest centroid. So the combination of assigning points to the nearest centroid and requiring a predefined number of clusters is characteristic of K-means clustering.

Partitioning data into a fixed number of groups by assigning each point to the nearest center is the hallmark of K-means clustering. In this method you decide how many clusters you want (the value of k) up front. The algorithm then alternates two steps: assign every point to the closest centroid, creating hard cluster memberships; and recompute each centroid as the mean of all points assigned to it. This process repeats until assignments stop changing, effectively minimizing the within-cluster variance for the chosen number of clusters.

This approach stands apart from other clustering methods. DBSCAN groups points based on density and can discover clusters of arbitrary shape without a predefined number of clusters; it relies on density parameters and can label points that don’t belong to any dense region as noise. Hierarchical clustering builds a tree of clusters and lets you choose how many clusters to keep by cutting the tree at some level, rather than fixing k at the start. Gaussian Mixture Models use probabilistic, soft assignments to components and model the data as mixtures of Gaussians, with cluster membership determined by posterior probabilities rather than the nearest centroid.

So the combination of assigning points to the nearest centroid and requiring a predefined number of clusters is characteristic of K-means clustering.