Predicting consumer's willingness to pay for sustainable food packaging in fresh-meal delivery sector: a machine learning approach

Purpose Growing consumer concern for sustainability has increased the adoption of sustainable packaging in last-mile fresh-meal delivery services. Despite this trend, limited research explains why consumers are willing to pay a premium for sustainable packaging. This study aims to predict consumers' willingness to pay (WTP) for sustainable packaging by examining the role of key psychographic and perception-based factors. Design/methodology/approach Survey data were collected from 808 consumers using last-mile fresh-meal delivery services. To address the ordinal nature of attitudinal data and improve predictive robustness, a collapsed Likert-scale approach was employed. Machine learning classification techniques (logistic regression, random forest, gradient boosting and support vector machines) were applied to model the relationship between WTP and eco-consciousness, price sensitivity, brand trust, perceived packaging sustainability, perceived packaging quality and perceived packaging innovation. Findings Gradient Boosting emerged as the best-performing model, achieving the highest predictive accuracy and macro-F1 score. The results reveal that price sensitivity and eco-consciousness are the most influential predictors of consumers' WTP for sustainable packaging. Perceived packaging sustainability and perceived packaging quality also play significant roles, while brand trust exerts a partial yet meaningful influence on premium payment decisions. Practical implications The findings provide actionable guidance for marketers and platform managers in the fresh-meal delivery sector. Emphasizing sustainability benefits alongside packaging quality, while addressing consumer price sensitivity, can enhance perceived value and support premium pricing strategies for sustainable packaging solutions. Originality/value This study advances consumer marketing research by combining psychographic drivers of sustainable consumption with machine learning-based prediction. It also demonstrates the value of collapsed Likert-scale data for modelling consumers' WTP, offering both methodological and managerial contributions to sustainability-driven marketing decisions.