Correlation of subway turnstile entries and COVID-19 incidence and deaths in New York City

Abstract

In this paper, we show a strong correlation between turnstile entries data of the New York City (NYC) subway provided by NYC Metropolitan Transport Authority and COVID-19 deaths and cases reported by the NYC Department of Health from March to May 2020. This correlation is obtained through linear regression and confirmed by the prediction of the number of deaths by a Long Short-Term Memory neural network. The correlation is more significant after considering incubation and symptomatic phases of this disease as experienced by people who died from it. We extend the analysis to each individual NYC borough. We also estimate the dates when the number of COVID-19 deaths and cases would approach zero by using the Auto-Regressive Integrated Moving Average model on the reported deaths and cases. We also backward forecast the dates when the first cases and deaths might have occurred.

Keywords

• COVID-19
• Time-series analysis
• New York city subway
• SARS-CoV-2
• Long short-term memory
• ARIMA