Journal of Emerging Trends in Engineering and Applied Sciences: Research Areas

Article Title: Discovering Metrics of Waste-to-Wealth and Renewable Energy for Sustainable Community Development: A Comparative Analysis of Localized versus Global Machine Learning Recommendation Systems
by A.H. Zaharaddeen

Abstract:
This study develops and validates a hybrid machine-learning recommendation framework for identifying optimal waste-to-wealth conversion technologies, comparing localized Nigerian data against global aggregated data. Two experiments were conducted: (1) a Nigerian dataset integrating national waste composition statistics (n=1,247 municipal samples) and energy demand profiles; and (2) a global dataset compiled from 37 countries. A hybrid architecture combining Random Forest prediction with Multi-Criteria Decision Analysis (MCDA) generated ranked technology recommendations. The localized dataset achieved 91.3% recommendation accuracy and 1.52 kg CO?-equivalent reduction per kg waste�significantly outperforming the global dataset's 87.4% accuracy and 1.36 kg CO? reduction (p<0.01). Findings establish that dataset localization substantially enhances circular-economy decision systems by preserving context-specific waste heterogeneity. The framework provides actionable guidance for policymakers seeking evidence-based waste-to-energy strategies.

Keywords: Machine Learning, Circular Economy, Nigeria, Sustainability, Dataset Localization

Download full paper