How Covid-19 impacted pedestrian and bike mobility in Zurich

Teresa Kubacka

v0.8, 2021-01-04 (work in progress)

Data Source:

  • data from automatic pedestrian and bike counters
  • measurements in most cases done in 2 directions; they are not always similar (e.g. Hardbr├╝cke)

Data preparation:

  • aggregate the data into 1-hr bins
  • pick only the id1 of the counters that have data for 2020 and at least a yearly baseline (otherwise we cannot reason about anomalies in 2020)
  • counters count either bikes or pedestrians - group them into 2 families
  • did not apply the correction factor, as it is the same for all the data from the same counter, and now we want to reason about trends; -> TODO
  • if not specified, both directions are summed up together
  • always normalization by the # of days per bin (bin e.g. a week in a given year, or an hour in a given weekday in a given year) -> counters are not always available

Mobility per day-of-the-week and time-of-the-day

There is a distinct daily pattern in mobility:

  • weekdays = 3 peaks (commute to work, have lunch, go back home)
  • weekends = long wide peak (afternoon stroll)
  • nights = some areas have nightly mobility comparable with Sunday afternoon mobility