2DSTAT: a machine learning-based system for predicting consumer sentiment and behaviour in supermarkets

Abstract

Understanding consumer behavior in supermarkets is vital for optimizing marketing, product placement, and inventory management. Retailers and manufacturers can better tailor offerings by analyzing consumer interactions [1],[2]. Traditional analytics focus on sales data, often missing the complexities of decision-making [3]. This study introduces 2DSTAT, a machine learning-based system that integrates qualitative behavior data with sales records. Using video processing and deep learning, it provides insights to help manufacturers optimize product placement, refine marketing, and enhance engagement, improving shopping experiences and boosting sales. Currently, the study focuses on ‘biscuit’ products due to their high demand across all age groups [4], with plans to expand to other supermarket items.