We all know that if you drive in rush hour, you will make slower progress than driving at a quieter time of day. But is the same true for a cycle commute? My data says yes – lots of cars slow you down even when you’re on a bike:
The graph plots the time of day I left home versus the average speed for the ride, plotted for my last 200 morning commutes on my standard route , a 24.2 mile ride from just outside the M25 right into the centre of London.
Though there is some noise, there is a clear trend – earlier is faster. Leaving before 6am, I tend to achieve a speed of ~18mph, whereas when I leave at 7am the speed is less than 16mph. In fact plotting the trendline shows a gradient of 2mph lost per hour. Phrased another way, if I leave the house at 6am, I expect to arrive at the office at 7.21am, but leaving at 7am I expect to arrive at 8.33am – a full twelve minutes slower.
What causes this delay? In my experience, it is traffic. Although as a cyclist you don’t have to follow the vehicle in front all that much – when cars are going slowly in a traffic jam, you can generally filter past them all – they still slow you up. Filtering is slower than open riding and busy routes means more time at traffic lights – especially when traffic lights are on the SCOOT system as in London. You also have to slow down out of caution – if you have five times the amount of traffic at 7am as 6am, you are likely to have five times the number of nutters.
But could there be other factors at play rather than just traffic? Here are some possibilities:
- When it is cold, dark or wet, I am likely to have trouble getting myself out of bed, so am likely to leave late. But also the cold, the dark and the wet will slow me down when cycling – this would great a downward slope on the graph even if traffic was constant
- If I’ve looked at the forecast and know that the wind is my favour, I’m likely to be keen to get up and go because the ride will be easier. Conversely knowing a headwind lies in wait might make me hit the snooze button for five minutes. Again this would introduce a correlation between departure time and speed that is independent of traffic. Similar if I am tired I’m likely to get up late and be slow.
It is hard to separate out all these factors. One thing we can do though is that there is a particular time of year when it tends to be cold, dark and wet: winter. Therefore it is useful to look at Summer (June-August) and Winter (December-February) rides separately and see if I leave later in Winter – which would lend weight to the theory 1)
Here are the regression results for winter:
and for summer:
So even controlling for conditions a little bit, we still see slower rides at later departure times. The dependence is less strong in summer (1.8mph/h) but even stronger in winter (3mph/h). I think this is possibly just statistical noise – each data set now only contains ~50 rides, but it is also possible that the difficult conditions of winter mean that heavy traffic makes riding even more difficult.
But do I leave later in winter? Here are the stats:
- Average departure time (all commutes) : 6.25am
- Average departure time (summer commutes) : 6.25am
- Average departure (winter commutes) : 6.24am
- Average speed (all commutes): 17.0mph
- Average speed (summer commutes): 17.6mph
- Average speed (winter commutes): 17.1mph
So, surprisingly to me, NO – I even manage to leave one minute earlier in winter. Not surprising the summer rides are a little quicker on average (17.6 versus 17.1), which is not surprisingly – more light and fewer clothes to lug around. NB the overall average is 17.0 so I must be really in slow in spring/autumn!
So the stats suggest that the “it’s cold so I’ll be late and slow” theory is not valid.
Even when you’re on a bike, the rush hour traffic slows you down.
How the data was obtained
All my commutes are on Strava and then shared with VeloViewer. Amongst VeloViewer’s many many little features is the ability to download a CSV file that contains summary data about all your rides. This file opens in Excel or similar spreadsheet program. From there it is a simple-ish job to filter down to the rides of interest.