This project explores unsupervised learning techniques to segment customers and identify associations between purchased products. Using a publicly available dataset containing demographic and transactional data, clustering algorithms (K-Means, DBSCAN, hierarchical clustering) and dimensionality reduction methods (PCA, UMAP) are applied to group customers based on behavioral patterns. Association rule mining (Apriori, Eclat) is employed to detect frequently co-purchased items. The results demonstrate the utility of these techniques for generating actionable insights without relying on predefined labels.