When data has no labels, an algorithm cannot learn from examples marked with the right answer. Unsupervised learning instead looks for patterns and structures in the data itself, often grouping items by their similarities and differences. Unlike supervised learning, it does not need someone to label each example first. section 1.1 section 3.6
Finding customer groups
A retailer can use clustering to group customers with similar purchasing behavior. The resulting groups might reveal budget-conscious shoppers, luxury enthusiasts, and occasional buyers. Retailers can use those segments to refine marketing, recommendations, and inventory decisions; the algorithm finds the groupings, while people must decide what they mean and how to use them. section 3.6
The approach is useful when labeling a large, high-quality dataset would be difficult or expensive. Patterns in unlabeled data can offer leads for decisions, including work against money laundering, but finding a pattern is not the same as proving what it means. For an AI project, consider unsupervised learning alongside other techniques and check that the resulting system works as expected. section 3.6 section 11.8 section 11.28
