Predictive Management of EV Charging Stations Using Machine Learning
Alex David S, Almas Begum, D Hemalatha, K Senthil, R Monikaa, Ruth Naveena N
Abstract
Based on experimental results the predictive accuracy of Random Forest, Decision Tree, and XGBoost was much higher than the Linear Regression algorithm in EV Charging Management using machine learning models. Random Forest handled the charging station dynamics well, and the complex data relationships were managed easily with XGBoost. On the other hand, Linear Regression performed better because results displayed high errors and its model fit numbers were not significant. In contrast, the Decision Tree has glaring issues with overfitting and generalization. The results highlight the importance of advanced data science methods for efficient EV infrastructure management by yielding exact forecasts about demand and energy consumption behaviors. Future endeavors will continue to build upon these models, increasing the scale of analysis and providing state-of-the-art solutions towards developing an approach for both scalable development as well as a push toward more sustainable transportation with improved operational efficiencies and reliability within EV networks.