Answers are grounded in this paper and cite their sources.
• Behavioral intention is the strongest predictor of household waste separation. • Gradient Boosting Regression achieved high accuracy (R 2 = 0.782) in behavior prediction. • Social norms and emotional drivers outweigh policy or demographic factors. • Community feedback had limited standalone impact but enabled targeted strategies. • A novel, data-driven framework aligns waste policy with ESG and SDG objectives. This study examines the behavioral and social factors influencing household waste separation practices in Keelek Subdistrict Municipality, Chiang Mai Province, Thailand. Drawing on survey data and a mixed‐methods approach that integrates correlation analysis, principal component analysis, and a two‐stage machine‐learning pipeline—further validated by confirmatory structural equation modeling of Attitude → Intention → Behavior and mapped onto an established nudge taxonomy—the research identifies the most influential predictors of separation behavior. These include routine organic waste sorting, behavioral intention, emotional commitment, and the perceived influence of community members and local authorities. Among the tested models, Gradient Boosting Regression yielded the highest predictive accuracy (R 2 = 0.782; MAE = 0.331), underscoring its ability to capture complex non-linear behavioral patterns more effectively than traditional approaches. By uniting behavioral theory, community-derived insights, and predictive analytics, this work advances a novel, transferable framework for municipal planning. It offers practical, ESG/SDG–aligned strategies—such as habit-based, peer-supported nudges and AI-powered monitoring systems—that local governments can adopt to design evidence-based waste policies. Focusing on a semi-urban context often overlooked in the literature, this study fills a critical methodological gap and charts a replicable pathway for scaling behaviorally informed waste-management interventions.
No takes yet. Share an insight, caveat, or question.
Pumas et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: