Published Paper Details:

Understanding customer behavior in online shop- ping environments plays a central role in shaping marketing strategies and building personalized user experiences. This paper presents a machine-learning-based framework that leverages clickstream behavioral data to perform customer segmentation and deliver targeted product recommendations. The study utilizes a publicly available e-commerce dataset from Kaggle, which contains detailed records of user interactions such as product browsing, add-to-cart actions, and purchase completions. From this dataset, a set of behavioral features is derived that collec- tively captures spending patterns, order frequency, engagement intensity, and sensitivity to discounted products. These features serve as inputs to the K-Means clustering algorithm, an unsuper- vised machine learning technique that partitions customers into groups sharing similar behavioral characteristics. To improve the interpretability of the resulting clusters, Principal Component Analysis (PCA) is applied to reduce the multi-dimensional feature space into three principal components representing spending and order intensity, engagement behavior, and discount sensitivity. The reduced representation enables effective visualization of customer clusters and supports analysis of inter-group behavioral differences. Leveraging the identified customer segments, the system generates segment-specific product recommendations and delivers them through an automated email notification pipeline. The proposed approach demonstrates how clickstream-based behavioral analytics, combined with unsupervised learning and dimensionality reduction, can support meaningful personalization and targeted marketing in e-commerce environments.