Medium-term public transit route ridership forecasting: what, how and why ? A case study in Lyon

摘要: Abstract. Demand forecasting is an essential task in many industries and the transportation sector no exception. This because accurate forecasts are a fundamental aspect of any rationale planning process component intelligent systems. In context public transit, needed to support different level organisational processes. Short-term forecast, typically few hours future, developed real-time operations. Long-term 5 years or more for strategic planning. Those two forecast horizons have been widely studied by academic community but surprisingly little research deal with between those ranges. The objective this paper therefore twofold. First, we proposed generic modelling approach next 365 days ridership transit network at levels spatiotemporal aggregation. Second, illustrate how such models can assist operators agencies monitoring supporting recurrent tactical tasks. formulation based on multiplicative decomposition that combines tree-based trend forecasting. evaluation unseen data proves generates coherent forecast. Different use cases then depicted. They demonstrate resulting various tasks as setting future goals, definition service provision. Overall, study contributes growing literature automated collection. It confirms sophisticated statistical methods help improve enhance data-driven decision making.