Predictive Analytics for EV Charging Demand: A Machine Learning Perspective

Efficient demand forecasting is critical for optimizing the operations of electric vehicle (EV) charging stations, ensuring grid stability, and enhancing user satisfaction. This paper presents a comparative analysis of five machine learning models—Decision Tree, Gradient Boosting, k-Nearest Neighbors (kNN), Support Vector Machine (SVM), and Random Forest (RF)—applied to demand forecasting for an alternating current (AC) connected EV charging station. The dataset is preprocessed to ensure reliability, with 70% of the data for training, 15% for model validating and 15% for testing. MATLAB was utilized for implementing and evaluating the machine learning models, providing an integrated environment for data analysis, model training, and performance visualization. Performance metrics were evaluated using root mean square error (RMSE), mean absolute error (MAE) and co-efficient of determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>) to compare the predictive capabilities of forecasting models. Among the models tested, Random Forest achieved the highest performance, with an RMSE, MAE and R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> value of 3.54kWh, 2.18kWh and 0.88, respectively, demonstrating its relative effectiveness in forecasting demand in this context. The findings highlight the potential of machine learning, supported by MATLAB's robust computational tools, in improving demand prediction for EV charging infrastructure.