A Data-Driven Machine Learning Approach to Identify End-of-Life Vehicles in Indonesia

This research addresses the issues of traffic congestion, economic losses, and air pollution in major Indonesian cities by focusing on End-of-Life Vehicles (ELVs) and emission standards. The study highlights the impact of rapid population growth and increased vehicle ownership on road infrastructure, resulting in prolonged commuting times and elevated air pollution levels. The research predicts emission thresholds and identifies non-compliant vehicles by mapping vehicles aged 10 to 30 years and employing machine learning techniques. Key findings reveal that 15-80% of gasoline vehicles and 40-70% of diesel vehicles fail to meet the 2023 emission standards. The study proposes emission thresholds of 245 ppm for hydrocarbons, 1.8% for carbon monoxide, and 70% for opacity in diesel vehicles, with a minimum age of 10 years for ELVs. Machine learning models, particularly XGBoost for gasoline vehicles and Gradient Boosting for diesel vehicles, demonstrated high accuracy in predicting non-compliant vehicles. Policy recommendations include stricter emission testing, financial incentives for retiring old vehicles, and establishing low-emission zones. Implementing these policies, supported by robust public awareness campaigns and stakeholder engagement, is expected to significantly reduce air pollution, improve public health, and enhance urban transportation systems. The research provides valuable insights for policymakers to develop sustainable transportation strategies and effectively manage ELVs, contributing to Indonesian cities' overall quality of life.