Optimizing Electric Vehicle Range Prediction Using Machine Learning: A Feature-Based Comparative Study

Yijia Yang

2025

Abstract

Accurate prediction of electric vehicle (EV) driving range is essential to addressing consumer range anxiety and improving energy planning. This study investigates a feature-based comparative approach to EV range prediction by integrating real-world vehicle specifications and battery characteristics. A cleaned dataset of 102 EV models from Kaggle was analysed using three machine learning algorithms: Light Gradient Boosting Machine (LightGBM), Extreme Gradient Boosting (XGBoost), and Gaussian Process Regression (GPR). Variables such as battery capacity, energy efficiency, fast charging rate, and top speed were selected based on their measurable correlation with EV range. Pearson correlation analysis and LightGBM feature importance visualization revealed Battery_Pack_Kwh and Efficiency_WhKm as dominant predictors. A linear regression model, implemented in R, achieved high predictive performance with an R² of 0.969 and MAE of 17.08 km on the test set. Residual diagnostics, Q-Q plots, and predicted-vs-actual comparisons confirmed the model’s reliability. The findings underscore the importance of data-driven modelling and suggest that even moderately correlated features can enhance prediction when modelled non-linearly.

Paper Citation

in Harvard Style

Yang Y. (2025). Optimizing Electric Vehicle Range Prediction Using Machine Learning: A Feature-Based Comparative Study. In Proceedings of the 2nd International Conference on Innovations in Applied Mathematics, Physics, and Astronomy - Volume 1: IAMPA; ISBN 978-989-758-774-0, SciTePress, pages 233-239. DOI: 10.5220/0013822700004708

in Bibtex Style

@conference{iampa25,
author={Yijia Yang},
title={Optimizing Electric Vehicle Range Prediction Using Machine Learning: A Feature-Based Comparative Study},
booktitle={Proceedings of the 2nd International Conference on Innovations in Applied Mathematics, Physics, and Astronomy - Volume 1: IAMPA},
year={2025},
pages={233-239},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013822700004708},
isbn={978-989-758-774-0},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Innovations in Applied Mathematics, Physics, and Astronomy - Volume 1: IAMPA
TI - Optimizing Electric Vehicle Range Prediction Using Machine Learning: A Feature-Based Comparative Study
SN - 978-989-758-774-0
AU - Yang Y.
PY - 2025
SP - 233
EP - 239
DO - 10.5220/0013822700004708
PB - SciTePress