Design of a Forecasting Method for Occupancy Rates in Local Public Transport Based on Data from Automatic Passenger Counting Systems

Abstract

The predominant focus on individual motorized transport is neither sustainable nor socially just. One goal of a more sustainable design of the transport sector is to encourage people to use public transport. One barrier for passengers to use public transport are heavily occupied vehicles and the uncertainty about whether an empty seat is available on the desired connection. In this paper, a model is presented that is able to forecast the occupancy of vehicles in public transport. This information can be provided to passengers to increase customer satisfaction. Different sub models are presented, which differ according to their forecast horizon and the data sources used. The most important data source is data from automatic passenger counting systems collected in vehicles in the region of Northern Hesse during the project period of the research project U-hoch-3. After linking further data sources such as weather and timetable data, stratification characteristics are developed based on which occupancy states can be derived for future journeys. By linking the data with real-time data, the forecast quality can be significantly improved. It is shown which influences the Covid-19 pandemic and the introduction of the 9 € ticket in Germany had on the model development and by which functions these changes in demand can be correctly represented by the model. The results presented in this paper show that it is possible to reliably predict occupancy rates for vehicles in public transport.