On 26th November 2022 I recorded my 150th BRC club run on Strava. A “Daddy Hundred” in the words of the great Graham Gooch. The ride itself was an uneventful but enjoyable jaunt out to Cheddar, with a long stop at Café Gorge. Nice.
I didn’t realise I’d reached this milestone at the time. No, I learnt this during a couple of quiet days between Christmas and New Year when – driven by boredom, curiosity and a desire to learn some data analysis skills – I decided to delve into my Strava data to see what it might reveal.
The aim was to look for insights into the club run generally. Where do we ride? How far do we ride? How fast? How much climbing? How do club runs vary through the seasons?
In many cases the data confirmed what I think we all strongly suspected (we go to Cheddar a lot, for example), but equally, there were a couple of interesting surprises.
From 2017 onwards, it turns out I’ve ridden roughly half of all club runs, 154 in total. By no means a complete set, but enough to get a good idea of what generally happens on a Saturday morning.
Most of these rides were with the ‘medium group’, although a significant minority – particularly in the earlier years – were with Brian’s group. On a couple of occasions, I completely lost my mind and went out with the fast group. Only ‘full’ club runs have been included – rides where I peeled-off early for whatever reason have been omitted.
The dip in the number of club runs in 2020 was of course because – for a fair chunk of this year – riding with people outside one’s household was illegal.
All 154 club runs are plotted on the heatmap below. The brighter the track, the more times that route has been ridden:
What does this tell us? Well, the main things that jump out at me are:
Note: the long route down south into Dorset was, I believe, one of Ali’s splendid all-day magical mystery tours in the summer of 2021.
It varies! The shortest ride in the sample was 73.7 km, while the longest was 158.7 km – just shy of 100 miles. However, the chart below reveals that the majority of rides were between 90 km and 120 km. The average was 106.2 km.
Again, it varies quite a bit. Average speeds ranged between a pedestrian 23.8 kph and a brisk 30.9 kph, though the chart below shows that the vast majority of club runs that I rode were ridden at average speeds between 25 kph and 29 kph. The overall average speed was 27.2 kph.
Note that the rides ridden at 30-31 kph were my two painful outings with the fast group mentioned earlier.
In the cold, wet winter months there are many potential reasons for lower average speeds on the bike. Bulky heavy clothing, heavier bodies, heavier bikes, slower winter tyres, denser air, rougher road surfaces, muscles not working as well in colder temperatures, stronger winds and a general lack of motivation are just some of the excuses I’ve heard bandied about over the years. But how much does all this really slow us down? The chart below shows the spread of my average speeds in each of the seasons.
At first glance, it appears that we slow down a lot in the winter, riding on average, about 2 kph slower than spring and autumn, and about 3 kph slower than summer. Almost all winter rides are slower than the slowest summer ride. However, I suspect this may be skewed by the fact that I probably rode with the slower group more often in the winter.
There’s also a lot of variation in average speed within the shoulder seasons of spring and autumn, perhaps reflecting the very different weather conditions at the beginning and end of these seasons.
The chart below shows the elevation gain of each ride in the sample, plotted against ride distance.
As you’d expect, longer rides generally have more elevation gain, compared to shorter rides. There are a few short hilly rides in there, along with a smattering of longer flat ones, but generally, we climb roughly 100 m for every 10 km of distance. In my book, this translates to “moderately hilly”.
A few people have expressed the opinion to me recently, often while struggling up a climb in the Wye Valley, that the medium group ride is getting harder. Does the data back this up? Or are these people just getting softer and/or less fit?
It does appear from the charts above that club runs in 2021 and 2022 were generally a little bit longer and hillier than previous years, though not by a huge amount. Average speeds in 2021 and 2022, although higher than 2020, were comparable to 2019.
So there you have it. With the caveat that there is a bias in the data towards the medium group ride, the myth that we only ever go to Cheddar or Monmouthshire has been well and truly busted. We’ve graced the roads of many of the surrounding areas of Bristol over the last few years, except the Bath area, which we’ve avoided like the plague.
There is some evidence that medium group rides have become slightly longer and hillier over the last couple of years. Slightly. Those that have found themselves a semi-conscious wreck unable to leave the sofa on a Saturday afternoon during this period can therefore console themselves in the knowledge that this may not have been entirely due to deteriorating fitness.
Fancy analysing your own data?
If you’re interested in doing something similar with your own Strava data, you’ll first need to get hold of your data (obviously). One way to do this (and what I did) is to request a full export of your data from Strava:
What you get will include absolutely everything; activity data, GPS tracks, created routes, comments on rides, kudos’s given and received, support tickets raised, uploaded media, profile details, etc etc. It can be enormous, particularly if you ride a lot and/or upload lots of photos and videos. I just used the activities.csv file and the accompanying GPS files for this analysis.
Alternatively, Strava provide an Application Programming Interface (API) that allows you to pull data from the site using your own application. I don’t believe you can get everything this way however, including GPS data. More information on the API can be found here:
Once you’ve got your data, choose a data analysis tool. I used Python, with its vast array of free Data Science libraries and large user community (=lots of online training/support), but there are plenty of other options. You could probably do a fair amount with Excel.