Comparative Analysis of Fuel Types on Fuel Efficiency and $\text{CO}_{2}$ Emissions in Light-Duty Vehicles with Machine Learning Prediction Models

Abstract

The emergent desire for eco-efficient transportation has increased concern about how fuel type influences fuel consumption and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{CO}_{2}$</tex> emissions in LDVs. This research examines the fuel economy and emissions characteristics of several fuel types, including gasoline, diesel, and E85 ethanol, using Canadian vehicle fuel consumption and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{CO}_{2}$</tex> emissions information from 2000 to 2022. The dataset obtained from Kaggle includes metrics specific to each model, which allows modeling real-life driving scenarios while comparing models is precise. Some findings include that diesel provides the lowest fuel consumption and emissions compared to other fuels while ethanol provides the highest fuel consumption and emissions figures. Given vehicle parameters, the prediction of fuel consumption was improved using several machine learning algorithms, out of which the Random Forest Regressor model demonstrated high accuracy, averaging an MSE of 0.00036 and an <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{R}^{2}$</tex> of 0.97686. This model exhibited the highest accuracy of data variation, making it a valuable prognosis instrument for fuel consumption. Information from this study benefits the consumer who intends to make green choices and the policymakers who want to make policies on vehicle emissions. In enhancing the performance of fuel consumption rate prediction, this paper advances knowledge in the design of more efficient vehicles and promotes efforts concerning environment-friendly transport.