Analysing the impact of weather on bus ridership using smart card data

Abstract

This study investigates the impact of weather on bus ridership using smart card data collected in Fengxian, Shanghai. The ridership data are categorised into three representative groups by the cluster analysis. The ridership data for each cluster are further divided according to the four seasons. Twelve separate multiple linear regression models with four weather variables and two dummy variables are constructed and calibrated. All four weather variables, namely humidity, wind speed, rainfall and temperature are found to have statistically significant negative effects on bus ridership. The magnitude of the impact varies depending on bus route types, seasons and mode share characteristics. Our analysis provides a valuable case study on weather's impact on bus ridership and concludes that there is no one-size-fits-all conclusion about the relationships between weather attributes and bus ridership, and it is critical to investigate those relationships in different geographical contexts. The results of this study can be used not only for long-term transit policy making but also as a decision making tool for short-term ridership forecasting.