Exploring sustainable consumer behaviour in retail: A teacher-student model framework for socio-economic insights

Abstract

This study examines the socio-economic factors that shape consumer behaviour related to carbon footprint awareness within the retail and services sectors. Utilizing a structured, dual-model framework, we analyse the complex relationships between key socio-economic determinants, such as income, education, age, gender, and environmental consciousness, and sustainable behaviours, including recycling, green product searches, and engagement with eco-friendly practices. The methodology combines two machine learning models to achieve both detailed analysis and interpretability. The primary model, based on the Tabular Network Regressor (TabNet), captures nuanced, high-dimensional interactions across socio-economic variables, while the secondary, interpretive model, using the Categorical Boosting Regressor (catBoost), simplifies these insights without compromising accuracy, making findings accessible for policy-oriented applications. Our results reveal that income, education, and environmental consciousness are significant predictors of carbon footprint awareness. Moreover, consumers who regularly engage in sustainable actions, such as recycling or searching for eco-friendly products, exhibit heightened awareness of their environmental impact, underscoring the influence of habitual eco-conscious behaviours. Through a Shapley Additive Explanations (SHAP) analysis, this study enhances the interpretability of results, providing a clear, data-driven understanding of how specific socio-economic factors contribute to sustainable consumer actions. This research advances knowledge on socio-economic influences in sustainability by offering actionable insights for policymakers, including targeted strategies that address diverse demographic characteristics and promote inclusive, effective policies for sustainable consumer behaviour in the retail and services industries.