Availability Status Prediction of EV Charging Stations via Deep Learning and Decision Tree Explainability

Abstract

The increasing adoption of electric vehicles (EVs) has intensified the need for efficient management of charging infrastructure. This project addresses the challenge of predicting the availability status of EV charging stations — classifying whether a slot is available or not available — using a deep learning model and enhancing its interpretability through a decision-tree- based Explainability approach. A sequential LSTM-GRU model was developed to predict the availability status of charging slots at multiple stations in Paris, incorporating temporal, spatial, and contextual features such as time of day, day of the week, location coordinates, and trend indicators. To ensure the transparency and reliability of the deep learning model’s predictions, a Decision Tree classifier was employed as an interpretable surrogate model. By analyzing the feature importance’s derived from the Decision Tree, the study identified ’Longitude’ and ’Latitude’ as the most significant factors influencing charger availability, highlighting a strong spatial dependency in EV infrastructure usage patterns. The integration of interpretable models alongside deep learning models enhances decision-making confidence and provides actionable insights for urban mobility planners and infrastructure managers.