Which clustering limitation means that K-means cannot reliably identify clusters that are not spherical in shape?

Get ready for the GARP Risk and AI Exam with flashcards and multiple choice questions. Each question comes with hints and explanations. Prepare for success!

K-means relies on distances to centroids and partitions space into regions around those centers, which tends to produce spherical, evenly sized groups. The algorithm minimizes within-cluster variance, so the decision boundaries form around the mean in a way that favors round, compact clusters. When clusters are non-spherical—elongated, crescent-shaped, or otherwise irregular—the method struggles to fit them with a single center. It may split a single non-spherical cluster into multiple roundish groups or merge distinct shapes, leading to poor clustering. So the limitation is the difficulty with non-spherical clusters. Spherical clusters are actually well-suited to K-means, making that option not a limitation. Overlapping clusters and hierarchical clustering describe other issues or methods, not the core shape limitation of K-means.