Forecasting Charging Point Occupancy Using Supervised Learning Algorithms

The prediction of charging point occupancy enables electric vehicle users to better plan their processes and thus promotes the acceptance electromobility. study uses Adaptive Charging Network data investigate a public workplace site for predicting individual station as well overall occupancy. Predicting is formulated classification problem, while total regression problem. effects different feature sets on predictions are investigated, whether model trained all points per performs than one specific point. Reviewed studies so far, however, have failed compare these two approaches benchmarks, use more algorithm, or consider site. Therefore, following supervised machine-learning algorithms were applied both tasks: linear logistic regression, k-nearest neighbor, random forest, XGBoost. Further, results compared three naïve which provide robust benchmark, training sites. By adding features, quality can be increased considerably, resulted in some models performing approaches. In general, perform slightly median points. certain cases, individually achieve best results, with very low relative benefit from that has been data.