Car-Specific Metro Train Crowding Prediction Based on Real-Time Load Data

The paper formulates the car-specific metro train crowding prediction problem based on real-time load data and evaluates the performance of several prediction methods (stepwise regression, lasso, and boosted tree ensembles). The problem is studied for multiple stations along a metro line in Stockholm, Sweden. Prediction accuracy is evaluated with respect to absolute passenger loads and predefined discrete crowding levels. When available, predictions with real-time load data significantly outperform historical averages, with accuracy improvements varying in magnitude across target stations and prediction horizons.