Multi-Channel E-commerce Customer Interaction Dataset

The dataset employed in this study, titled “Multi-Channel E-commerce Customer Interaction Dataset,” consists of real-world data collected over a 12-month period from a major online retail platform. It integrates more than 2.3 million records encompassing 150,000 customers, 25,000 products, and multiple interaction types across five primary sources: browsing behavior logs, transaction histories, customer demographic profiles, product characteristics, and social interactions. Browsing logs capture page views and session durations, transaction data provide purchase and payment details, demographic profiles include static attributes such as age and location, product characteristics cover categories, pricing, and ratings, while social interactions represent reviews and peer recommendations. The dataset is anonymized to comply with privacy regulations, and preprocessing steps such as imputation and feature engineering were applied to ensure data completeness and to construct graph-relevant representations for modeling customer-product relationships.