01698062

Component

This study investigates the effects of different variables including meteorology, trip characteristics (such as time of day), driving characteristics (such as the frequency of extended idling), and driver characteristics (such as driving experience) on trip-level emission factors (EFs). 82 participants were recruited to collect their driving activities for a one-week study period from March to July 2018 in the Greater Toronto Area (GTA). Totally, 1,113 driving trips were collected. A database was created for estimating emissions by generating and organizing 51 independent variables for each trip. An eco-score evaluation system was established based on log-transformed emissions of greenhouse gases (GHG) in CO₂eq. A machine learning approach, the Extreme Gradient Boosting (XGBoost), was used to develop prediction models for CO₂eq emissions at a trip level. The coefficient of determination (R²) and root-mean-square-error (RMSE) for the model were 0.84 (std. dev. 0.05), and 10.26 (std. dev. 1.24). The Shapley additive explanation (SHAP) measures were employed to reveal the importance of various features affecting trip emissions. The analysis was conducted in two stages: First, including all independent variables with the exception of variables typically used in emission estimation models, and second solely including discrete variables representing driver characteristics, meteorology, vehicle characteristics, and trip characteristics. Continuous features associated with driving behavior were found to have the most significant impact on the trip eco-score. Additionally, driving experience was the most significant discrete feature affecting the eco-score. Finally, commuter drivers were found to have lower emission intensities, probably because they are more familiar with their route.

This paper was sponsored by TRB committee ADC20 Standing Committee on Transportation and Air Quality. Alternate title: A machine learning approach capturing the effects of meteorology, time of day, driving behaviour, and driver experience on trip-level emissions.

19-03670

English

22p

2019

Transportation Research Board 98th Annual Meeting

Digital/other

Figures; Maps; References; Tables

Automobile drivers; Automobile driving; Driver performance; Ecodriving; Environmental impacts; Exhaust gases; Greenhouse gases; Machine learning

Real world data

Toronto Metropolitan Area

Environment; Highways; Safety and Human Factors; Vehicles and Equipment

Transportation Research Board Annual Meeting 2019 Paper #19-03670

TRIS, TRB, ATRI

Dec 7 2018 9:45AM