Impact of the COVID-19 Vaccination on Bike Sharing Demand in NYC: ARIMAX and Bayesian Time Series Analyses

Abstract

The objective of this study is to investigate the impact of the COVID-19 vaccine on bike sharing demand in New York City (NYC). The vaccination’s impact was also compared with other known influences (COVID-19 cases counts, COVID-19 deaths, weather data, trip purpose, and more) to help improve bike share demand modeling in a pandemic setting. Autoregressive integrated moving average (ARIMAX) time series models were estimated for Brooklyn and Manhattan in both the pre-vaccine and post-vaccine periods. Mean absolute percentage error (MAPE) was used for model evaluation, and the average MAPE was below 5, suggesting a high-level accuracy in modeling. The Bayesian time series analysis results were very similar to the ARIMAX model results (with the exception of some parameters being significantly underestimated). The results of the time series analyses showed that vaccination did not have a significant effect on bike sharing demand in Manhattan but had a significant effect leading to increases in bike sharing demand in Brooklyn. Despite this, vaccination was not the main influence on bike sharing demand. For instance, in Brooklyn, grocery and retail/recreational shopping were strong influences both pre- and post-vaccine, indicating that regardless of vaccination, shopping is essential. COVID-19 cases counts had opposing effects in Brooklyn and Manhattan. Other findings include higher temperatures leading to increased bike demand, and precipitation and stronger winds leading to decreased bike sharing demand.