A Comparative Machine Learning Approach for Energy Consumption Prediction in Hybrid Vehicles
With the rise in environmental concerns and global shift toward sustainable mobility, accurately predicting energy consumption in hybrid vehicles (HVs) is essential for reliable transportation planning and optimizing efficient energy management. This study employs the National Household Travel Survey (NHTS) dataset to develop predictive models for estimating energy usage in HVs. In this paper, four machine learning models (XGBoost, Linear Regression, Random Forest, and Neural Network) are implemented to analyze and compare their effectiveness in forecasting energy consumption. The dataset is preprocessed and standardized, followed by model training and 5-fold cross-validation. The performance of each model is determined using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and the Coefficient of Determination (R²). Among them, the Random Forest model achieved the best effectiveness with lower RMSE and MAE along with the highest R2 score for predicting energy consumption in hybrid vehicles. The findings show the prospective of machine learning techniques, particularly the Random Forest model, which delivers the highest accuracy in predicting hybrid vehicle energy consumption. This result supports more informed decision-making for developing eco-friendly transport solutions, reducing fuel consumption and emissions.