A retailer has three years of purchase history for its loyalty members but has never assigned any of them to a segment or category. A data scientist wants an algorithm to examine the purchase patterns and discover natural groupings of similar shoppers on its own, without being told in advance what the groups should be. Which type of machine learning task is this?
Select an answer to reveal the explanation.
Short Explanation
The giveaway here is that nobody has ever labeled these customers with anything, there's no existing list of segments to sort people into. That rules out any approach that needs examples to learn from ahead of time, because there simply aren't any. What's left is an approach that looks at the raw purchase patterns and figures out on its own which shoppers behave alike, clustering them together without anyone telling it what the clusters should be called. Predicting a number, like expected next-month spend, is a different job entirely since it outputs a value rather than a grouping. And anything about images or physical objects has nothing to do with rows of purchase history. When the goal is 'find the hidden groups' rather than 'predict a label or a number,' that's the unsupervised, group-discovery style of learning.
Full Explanation
The correct answer is B. Clustering is an unsupervised learning task that groups data points by similarity without relying on predefined labels, which matches a retailer that has no existing segment assignments and wants the algorithm to find structure on its own. Option A is incorrect because classification is supervised learning that assigns new data to categories the model was already trained on using labeled examples, but here no labels exist yet. Option C is incorrect because regression predicts a continuous numeric value, such as forecasted spend, rather than discovering groupings among customers. Option D is incorrect because object detection identifies and locates items within images, which has nothing to do with analyzing tabular purchase history to find customer segments. Because the retailer wants unlabeled data organized into meaningful groups rather than sorted into known categories or used to predict a number, clustering is the appropriate approach.