Unraveling nonlinear and interaction effects of various determinants on bus gaseous emissions

Urban transit buses equipped with large-displacement engines operate on circular routes several times throughout the day, emitting large amounts of environmentally hazardous exhaust. Hence, understanding the intricate associations between bus emissions and multiple contributors is beneficial for creating sustainable transportation systems, while previous studies focusing on statistical methods fail to unravel them. This paper innovatively leverages the bagged decision tree approach to delineate such complex relationships based on the data collected from CNG-fueled and diesel-powered buses. Relative importance indicates that velocity appears to be the primary factor and is therefore selected as the research objective. Results suggest that the effects of different contributors on bus emissions present nonlinear patterns. More specifically, the influence of speed on CO, CO₂, and NOx exhaust generally reveals an increasing-stabilizing tendency while that of HC represents a decreasing-stabilizing mode. Besides, the phenomenon of synergies between determinants is also prevalent, for instance, buses within high-speed and large-slope conditions tend to produce more emissions. These findings can provide nuanced guidance for policy-making and bus route planning issues in consideration of environmental protection and pollution mitigation.