The Changing Dynamics of Dublin Bikes in 2020
July 27, 2020 by Philip Lowney

In an earlier post, the impact of the Coronavirus pandemic on the usage of the Dublin Bikes scheme was outlined in some detail. Another two months have passed, during which I have worked on some techniques to better visualise change over time using the Metrics Map and Day Profiles tools - both of which are available for everyone to burrow into the data for themselves. In this post, I'll try to shed some more light on how Dublin's scheme has been changing in 2020 thus far.

Pre-Pandemic

In old posts on the Dublin Bikeshare Scheme I outlined how each weekday, bikes flow from the north-west to the south-east quadrants of the scheme, with the same pattern repeated each evening. I created this video in 2018, and it is still a very fair summary of how the scheme used to behave before the Pandemic.

Note the washing back and forth of bikes each weekday, the emptying of stations rapidly each morning, and the rapidly filling stations on other opposite side of the city. This was a scheme in huge demand, getting people to work daily.

Weekday Flows - A Collapse with Signs of Recovery

The Morning Commute - 7 AM to 10 AM

The Metrics Map is a good tool to demonstrate the evolution of commuting patterns over time. The following GIF is a plot of the average net change in bikes per station from 7AM to 10AM, Monday to Friday. Each frame is one week's average values, starting the first week in January and finishing at date of publication. Red stations are gaining bikes, and blue ones are losing them. The animation pauses between loops, and the week being plotted is in the top-left.

I think there are three main phases here:

  • Until mid-March (when inital measure were introduced with the closing of schools etc.), the classic morning commuting pattern was very much in evidence.
  • In mid-March, usage practically halted, though the scheme remained open.
  • Since mid-May, we can see the gradual 'warming' of the scheme once again, with a greatly-attenuated version of the original trend emerging.

The Evening Commute - 4 PM to 7 PM

Let's perform the same exercise but this time we'll plot the data for the evening commute:

We see the same pattern as the morning commute, but in reverse. This symmetry suggests the scheme remains used for commuting, but simply to a lesser degree than before. For the figures on bikes in circulation, we must examine the Day Profiles.

A Note from the Author

I really hope you find the tools and blog posts provided here useful. This website is a hobby I undertake in my spare time. If you have enjoyed using this site, please consider sharing it on a platform of your choice using the social buttons provided. Knowing this project is useful to people spurs me on to provide further functionality and content. If you have any recommendations, please don’t hesitate to get in touch via the feedback link above.

Philip

Day Profiles - Evenings Now Busier

As data flows in to SchemeStats, we continuously calculate various metrics - number of slots in use, an estimate of the bikes in circulation, number of empty/full stations etc. One application of this data is in the generation of Day Profiles for these values, which are averages of the metrics over the course of the day, for a given day of the week during a range of calendar dates. There's a whole tool on this site dedicated to presenting this data, if that sounds interesting.

The following is an animation of weekly day profiles of numbers of bikes in circulation (estimated), for every week since the start of the year until the date of publication, with each frame capturing a different week. Three trendlines are included:

  • Saturdays and Sundays - these are the recorded figures for the weekend. This is the 'fuzziest' data, because there is only one sample point per week. Those weeks with multiple Saturdays are hard to come by, unfortunately.
  • Weekdays - these are the average values for Monday-Friday over the course of the week. Weekdays have typically behaved almost identically traditionally, as distinct from weekends, so it makes sense to plot them together.

I regret that a bug currently means the data for the last week is truncated. Time constraints mean I am offering this as-is.

Some observations of the Day Profile data:

  • Weekday usage collapsed in-line with the data indicated by the Metrics Map animations, with both morning and lunchtime usage remaining greatly reduced .
  • Weekday usage has recovered most strongly in the evenings, with peak usage recently approaching approximately 200 bikes in circulation by 6PM, a significant share of the former peak of 300-350 at a similar time.
  • Weekend usage has occasionally been extremely busy, rivalling any pre-coronavirus weekday in terms of bikes in use in the evenings.

Aside from the observations above, two thoughts occur:

  • Might the stronger recovery in evening figures compared to morning figures be attributed to greater leisure usage than before? Less people are commuting to the office, but perhaps some former commuters are taking the bike for a spin in the evening to get a some exercise?
  • Might the instances of extremely busy weekends we saw mid-pandemic be attributable to the now-removed restrictions on movement, with greater numbers of people remaining in Dublin and using the bikes on Saturday and Sunday afternoons?

Availabilty of Bikes & Slots is Much Improved

For those still using the scheme, it has become much easier to get a bike or a slot than before. Here is the evolution of Day Profiles for empty stations since January:

We can observe the figures of empty stations used to coincide with the end of the old commuting periods, the end of which has seen a likewise improvement in the figures.

The same is true for figures for full stations, of which there has been few, especially during the week:

Finally – stay safe, keep your distance, and don’t forget to wash your hands!