Comparative Machine Learning Models for Predicting Sustainable Food Purchase Decisions Using Eco-Labeling and Green Product Perception Analysis

This study investigates the determinants of green purchase decisions by integrating eco-labeling, green product perception, and purchase intention within a machine learning framework. The research compares the predictive capabilities of multiple linear regression and Random Forest models using survey data from 200 consumers of a leading green product brand in Indonesia. Data preprocessing, including logtransformation, was employed to mitigate skewness and enhance model robustness. Results indicate that eco-labeling exerts the most significant influence on purchase decisions, followed by purchase intention and green product perception. Linear regression demonstrated slightly superior predictive performance to Random Forest, suggesting that the examined variables predominantly exhibit linear relationships. The findings highlight eco-labeling's strategic role in bridging environmental attitudes and purchasing behavior and provide practical insights for marketers and policymakers aiming to promote sustainable consumption. This research also underscores the value of evaluating modeling techniques relative to data characteristics when studying green consumer behavior.