EV CHARGEWISE Predicting EV Charging Station Waiting Time

The research paper "EV CHARGEWISE: Predicting EV Charging Station Waiting Time" by Vaishnav Nakhate evaluates whether supervised machine learning models can predict electric vehicle queueing times using a dataset of 8,354 charging sessions across 20 stations. The author trained and tested Multiple Linear Regression, Ridge, Lasso, Polynomial Regression, Random Forest, Gradient Boosting, and a Multi-Layer Perceptron neural network against a simple mean baseline using features like station ID, vehicle type, location, battery capacity, and arrival time. The primary finding was that the mean baseline outperformed all machine learning alternatives—recording the lowest test RMSE of 4.797 minutes and proving that the available static arrival and station features lacked sufficient explanatory signal to predict session-level wait times. Despite this limitation, the study successfully demonstrated a complete end-to-end data science pipeline by deploying an interactive Streamlit application for predictions and station comparisons, concluding that accurate real-world queue forecasting requires dynamic operational variables like live charger occupancy and real-time queue lengths.