Research Article
The Day-Ahead Forecasting of the Passenger Occupancy in Public Transportation by Using Machine Learning
@INPROCEEDINGS{10.1007/978-3-030-97603-3_1, author={Atilla Altıntaş and Lars Davidson and Giannis Kostaras and Maycel Isaac}, title={The Day-Ahead Forecasting of the Passenger Occupancy in Public Transportation by Using Machine Learning}, proceedings={Intelligent Transport Systems. 5th EAI International Conference, INTSYS 2021, Virtual Event, November 24-26, 2021, Proceedings}, proceedings_a={INTSYS}, year={2022}, month={3}, keywords={Artifical intelligence SVR Machine learning Forecasting Public transport}, doi={10.1007/978-3-030-97603-3_1} }- Atilla Altıntaş
Lars Davidson
Giannis Kostaras
Maycel Isaac
Year: 2022
The Day-Ahead Forecasting of the Passenger Occupancy in Public Transportation by Using Machine Learning
INTSYS
Springer
DOI: 10.1007/978-3-030-97603-3_1
Abstract
Public transport is one of the main infrastructures of a sustainable city. For this reason, there are many studies on public transportation which mostly answer the question of “when my next bus will arrive?”. However now when the public is under the restrictions of the Covid-19 pandemic and learning to live with new social rules such as “social distance” a new yet crucial question arise on public transportation: “how crowded my next bus will be?” To prevent the crowdedness in public transportation the traffic regulators need to forecast the number of passengers the day ahead. In this study, in cooperation with Synteda, we suggest a machine learning algorithm that forecasts the occupancy in a bus or tram the day ahead for each stop for a route. The input data is past passenger travel data provided by the Västtrafik AB which is the public transportation company in Gothenburg, Sweden. The hourly data for the precipitation and temperature also has been added to the forecasting method; the database of precipitation and temperature is obtained by the SMHI, Swedish Meteorological and Hydrological Institute.
