Forecasting Daily High-Speed Rail Demand for Sustainable and Low-Carbon Mobility

Abstract. In light of the shift in Africa's transportation systems towards resilient and carbon-free transportation systems, Moroccan High-Speed Railway (HSR) qualifies as a green transportation alternative to transportation systems run on fossil fuels. It further represents a major shift in Morocco's decarbonization strategy. However, for such environmental advantages to be realized, there needs to be precise forecasting of the short-term demand for HSR transportation. This further leads to the fitting of the transportation schedules according to the passenger demand. In other words, it affects the consumption of traction energy. With the help of the daily passenger traffic collected from ONCF, the forecasting model for HSR transportation systems is established in this manuscript. This is an assessment of both traditional time series forecasting models and machine learning models. Both forecasting models capture the recurring patterns in HSR transportation systems according to the calendar. The results show that machine learning models better capture recurring demand patterns. In particular, the XGBoost model achieves the best performance, with error reductions of 25.6% and 22.9% compared to SARIMA, and 28.1% and 25.8% compared to SVR, confirming the potential of advanced forecasting models to support energy-efficient HSR operations in Morocco.