Interactive K-Means Clustering App for Retail Customer Data

Segmenting customers is often a retail analyst's first encounter with unsupervised learning, and the fastest way to understand K-Means is to run it on real shopper data and look hard at the results. This project keeps the focus exactly there: clean data in, visible clusters out, no distractions. The intern builds a Streamlit application where users upload a customer dataset, select the features to cluster on, choose a value of k, and run K-Means from scikit-learn. The app renders interactive scatter plots with points colored by assigned cluster and centroids clearly marked, so users can immediately judge whether the groupings look coherent, then re-run with different k values or feature choices to see how the segments shift. Python handles the fitting and plotting pipeline behind a deliberately simple interface that keeps the algorithm at the center of attention. The project gives the intern a solid grounding in centroid-based clustering — how initialization, feature choice, and k shape the outcome — along with their first experience shipping an interactive machine learning tool, making it an ideal proving ground before tackling segmentation work with formal validation techniques.