Publication

Abstract : E-commerce, short for electronic commerce, is the practice of placing orders online or negotiating a sale's terms and price. Although e-commerce is often associated with selling products and services to consumers, business-to-business (B2B) transactions also make use of it. To market their goods, many smaller consumer enterprises rely only on online sales. Nowadays, people are choosing to make purchases online as online transaction mechanisms and shopping platforms develop. Nevertheless, given that both clients and Merchants are unable to interact with clients face-to-face, there is a limited understanding of consumer demands. Due to this, businesses may struggle to gauge nuanced cues such as body language and personal preferences. Limited direct engagement can hinder the ability to address specific customer needs and preferences, potentially resulting in misinterpretation or overlooking crucial details. In turn, businesses may face difficulties in tailoring their products or services to meet the nuanced expectations of their customers, emphasizing the importance of alternative strategies such as detailed customer feedback mechanisms and data analytics to bridge this gap. The online system logs user activity and gathers data on consumer behaviour, allowing for the prediction of consumer purchasing patterns. Therefore, to improve the shopping experience of people, our study aims to use historical sales data to analyze consumer behaviour and do customer segmentation using RFM (Recency, Frequency, and Monetary) based K-means clustering and purchase prediction using various ML models to improve sales, out of which the combination of Logistic Regression and XGBoost performed the best.