Sports-associated events account for 8/10 (maybe 9?) top-ridership dates for BART (list and links to articles citing events and dates are below chart). I still have not (conclusively) determined catalyst for #10 on Oct. 2, 2019; I’ll be looking more closely at the by-station counts to determine stations of interest and further investigate – if you have any tips, let me know! However, ridership (~188k) on June 20, 2022 for the Warrior’s 2022 championship parade (first one since move Chase Center in SF) DID NOT make Top 10, barely on the scale of mean Saturday ridership (189k) pre-shutdowns.
The daily data compiled are from bart.gov, where hourly exit counts at each stations are summed for total count. Further visualization of long-term trends is presented as well, including precipitous drop and extremely low ridership since 2020.
Check out my GitHub for this project for code used to request, clean, plot data (including ridership quartiles by day of week) and preliminary forecast models (I used Facebook’s Prophet, Linkedin’s Greykite and standard ARIMA; as a true time-series model, the changes were better captured. Prophet and Greykite are built as standard linear regerssors parameters and coefficients to model holidays, seasons, weekends, etc.
Last updated June 2023
TOP 10 Ridership Days
- Oct 31, 2012 SF Giants Parade & Halloween
- Jun 16, 2015 Warriors Parade, Oakland
- Feb 5, 2016 Super Bowl L Village
- Jun 15, 2017 Warriors Parade, Oakland
- Oct 31, 2014 SF Giants Parade & Halloween
- June 16 2018 Warriors Parade, Oakland
- Feb 2, 2016 Super Bowl L Village
- Aug 29, 2013 Bay Bridge multi-day closure for (new) eastern span
- Feb 3, 2016 Super Bowl L Village
- Oct 2, 2019 ??? Oakland A’s AML wildcard Game 6
However, BART ridership was declining prior to COVID, since about 2018: click and hold to select a section in which to zoom for more detail on trends.The spike in mid-2022 is indeed ridership on June 20, 2022, the day of the parade, but the far below even weekend ridership before March 2020.
Ridership means in March 2022 were well below even 50% of pre-COVID.
The violin plots show the distribution pre-covid; hover-over show max, min, median and quartiles.
The following shows distributions for ridership since March 2020:
The plot above was published directly to chart studio from python notebook, thus is hosted there and I no longer have to embed long strings of HTML to generate within wp.
In my next post I’ll present models which look at fuel prices and car ownership as potential reasons for ridership decline, as well as map short vs long-distance (commute) trips to assess if there are differences in COVID-realted ridership for intra- vs inter- city trips.
