Uncovering E-Customer Insights through K-Means Clustering

Abstract

This research explores the dynamics of ecommerce consumer behaviour to help businesses understand their customers better. Instead of relying on old-school methods like  grouping people by age, location, or gender, we use a machine learning technique called K-Means Clustering. This approach looks at real customer behaviour—like how much money they spend (Sales), how much profit they generate, the discounts they use, and how many items they buy (Quantity)—to sort customers into meaningful groups. The best part? It doesn’t need any personal details about customers, so it works for almost any business, whether they’re selling clothes, gadgets, or home goods. 
The idea here to help companies make good decisions. By analysing these behaviour-based groups, businesses can craft personalized marketing campaigns, figure out how to keep customers loyal, and even tweak day-to-day operations to boost operational effectiveness. For example, businesses might recognize a segment of buyers driven primarily by promotional offers. Identifying such clusters allows companies to fine-tune discount strategies without compromising overall profitability. 
To democratize data-driven decision-making beyond technical experts’ data scientists), we also built a user-friendly, interactive dashboard using a tool called Streamlit. Imagine clicking through charts and graphs that show customer patterns in plain language—no coding required! This means managers, marketing teams, and even store owners can explore the data, ask questions like “Who are our most profitable customers?” and turn answers into action plans.