Use of Strava and PCT data in planning routes for cycle lanes and superhighways

17 posts / 0 new
Last post
landscape_urbanism
Use of Strava and PCT data in planning routes for cycle lanes and superhighways

The planning of cycle routes in London has often been based on the principle of looking for 'quiet' roads. Too often, these are on quiet backstreets which are probably quiet because of their indirectness. This seems a bad principle. The post-2015 Cycle Superhighways are based on the much-better principle of building segregated cycle lanes on strategic roads that provide direct origin-to-destination links. When it can be done this is surely the right thing to do.

To avoid spending money on unpopular cycle facilities data on cycle flows is necessary. An opportunity to see what routes cyclists use has come from the Strava Metro Global Heatmap. The problem with this data is that comes only from users of the Strava App. They tend to be young and young-at-heart male cyclists with a zest for competition. I have read that they may be 10% of  all commuter cyclists which would be a good-sized sample if not a representative sample. So I'd like to ask forum members to have a look at the Strava map for their local area and comment on whether it seems to be a fair representation of cyclists' route preferences. 

Here is a blog post which uses Strava data to comment on the intended location of the £65m Brunel Bike Bridge in East London. It was approved by Mayor Sadiq Khan in October 2016. It would be a pity if it was not built in the best location and it seems obvious that the location should be decided with reference to accurate historic data on cycle movement patterns. Is Rotherhithe the best place for the Brunel Bike Bridge? The bridge location was a Sustrans’ idea, not a TfL idea.

Tfl collects some cycling data with its (inevitably limited) cycle census operations by observers on street corners. The DT has sponsored a Potential to Cycle PCT tool which uses data from the 2011 national census and assigns it to cycle routes shown on the Openstreetmap. The tool was designed to assess potential, not for route selection.

Adler55

I lived in London (Hammersmith W6) for 6yrs from 2004 to 2010 and cycled for most of the time. I used all the back street routes *maps devised by Quil Forbes. I never had issues apart from the odd person complaining about me cycling the wrong way up or down a one-way street. Am so glad they're finally starting to create 'cycle only lanes' I lived in Germany for a few years and got spoilt with their wonderful cycling system. I basically continued in that fashion over here and have got away with it so far. Please keep pushing for cycling exclusive lanes with kerbs between cyclist & motorised vehicles. 

 

Cliff Matthews

Tried the link but it doesn't work, however I accessed the site by truncating the address provided by the link.

I looked at Ipswich. As a Strava user I can see that the data is heavily skewed towards well equipped cyclists. The A1214 figures heavily on the heatmap whilst activity in the centre of Ipswich is much more subdued, when in reality the weight of cycle activity I see is much greater in the centre but of course very few of these riders (including me) are recording their ride.

However it does give some useful information, particularly on a section of dedicated cycle route which the local labour Party want to open up to vehicular traffic.

Cliff Matthews

Tried the link but it doesn't work, however I accessed the site by truncating the address provided by the link.

I looked at Ipswich. As a Strava user I can see that the data is heavily skewed towards well equipped cyclists. The A1214 figures heavily on the heatmap whilst activity in the centre of Ipswich is much more subdued, when in reality the weight of cycle activity I see is much greater in the centre but of course very few of these riders (including me) are recording their ride.

However it does give some useful information, particularly on a section of dedicated cycle route which the local labour Party want to open up to vehicular traffic.

Old crank

I cerrtainly agree that Strava data, and the like, produces skewed data towards the 'athletic type' of cyclist. I have asked people I know, and others, to Strava all their rides reguardless of the type. Including the 2 mile round trip to the store for milk and bread. 

I also  started making my own segments which would reflect users who weren't likely to be going for KOM. Longer routes that I found usefull for commutes. Not the I was fasted uo the tenth mile hill. This I would use to inform decision makers that yes indeed people do cycle on these stretches of road and it would be good to make some better accomodation.

 

Cliff Matthews

Looking at the map, the skew effect is extremely heavy, probably due to the number of performance rides compared to recorded "ordinary" rides. So for example a road which I know is used fairly often (by me and my children) and others I have seen, it doesn't appear as having any bike traffic on it. There must be a high threshold number of rides before a section of road begins to register. I see a park which is crossed by a fair number of cyclists daily shows no colour, the adjacent road takes more cyclists than the busy dual carriageway but shows up as less busy.

What this map doesn't show is where people want to go but traffic danger prevents them.

mjray

I agree. The Strava map for King's Lynn suggests that some horrible stretches of A road have more cyclists on than some pretty good parts of National Cycle Route 1, which they definitely don't. About the only useful thing is that the lopsided colouring of one side of some roads which have fairly smooth cycle tracks on one side only suggests that some people using Strava are using the cycle tracks - could be an indication of tracks being good enough for half-decent speeds.

Strava is always going to be more popular among racers than shoppers and commuters. I also don't think it's good for cyclists to be unpaid recruiters for a private company.

Old crank

I do understand the qualms over cyclists being unpaid recruiters for a private company. (I use just the free version) But if you are already using the service it only makes sense to add to the data by including any other rides. In the US Strava is sellign the data to jurisdictions. This includes dates and times so traffic engineers/deciders can see what might be good locations to make changes and add infrastructure improvements. Yes there could be the posibility of abuse in personal data. I'm not privy to how Strava strips that data from the data set sent out. 

Even with its flaws, it is better data than the once a year at certain intersections that we normally get in the US.......

 

 

Old crank

I do understand the qualms over cyclists being unpaid recruiters for a private company. (I use just the free version) But if you are already using the service it only makes sense to add to the data by including any other rides. In the US Strava is sellign the data to jurisdictions. This includes dates and times so traffic engineers/deciders can see what might be good locations to make changes and add infrastructure improvements. Yes there could be the posibility of abuse in personal data. I'm not privy to how Strava strips that data from the data set sent out. 

Even with its flaws, it is better data than the once a year at certain intersections that we normally get in the US.......

 

 

landscape_urbanism

Sorry about the problem with the links in my post about data from the Strava App. Here they are:

http://labs.strava.com/heatmap/#6/-120.90000/38.36000/blue/bike

http://www.landscapearchitecture.org.uk/is-rotherhithe-the-best-place-fo...

landscape_urbanism

Thank you for the comments on the Strava App. It's inevitable that an enthusiast app shows enthusiast routes. But the idea of using mobile phones to log and store cycling data seems very useful. Cyclists need help from a coder to write an app which (1) can be made available as a competely free app (2) provides us with something so useful that we keep it switched on (3) sends data on journeys to an open source database. An organisation like Transport for London or the London Cycling Campaign could fund the app development. Any ideas for what the 'something so useful that cyclists keep it switched on' might be? Weather and time info don't seem enough.

Would you like an app which relates weather conditions to your personal journey times and gives predictions for when you will arrive at your destination? Maybe TfL could produce a cycle trip planner app. The TfL Journey Planner works on a mobile but even if .01% of cyclists used it they would not do so more than once.

michaelrobinson

TfL do use an app to collect cycling journey data - I was contacted to use it last year.  However, I think they use it to research specific projects rather than general data collection all the time.  The app developer was Mott McDonald, the engineering consultancy but it was totally targetted at collection of data for TfL and it wasn't a more general purpose GPS app like Strava or RideWithGPS.

SpacePootler

The trouble I have with any sort of GPS-based collection is that most people have an exclusion zone set around their home, to prevent being tracked down by malicious persons when the data are published online.

Many people set a zone of 1km or so, causing the genuinely short rides to be removed in their entirety, but most importantly removing the links these people use every single time they get on their bikes.  If we are to collect data, it should be done through anonymous aggregate analysis of routes themselves.

michaelrobinson

Last year I was contacted by TfL to install and use a smart phone app for GPS-based data collection of the routes I used for cycling in my area of West London.  I think I was on their system because I had previously responded to a consultation and they were using the data to look at options for a cycle superhighway route in the area and understand where people were cycling already.

I think the use of Strava data is fine provided the planners recognise the limitations and Strava data is better than no data and guesswork.

Where I live, I see the lycra warriors on their carbon bikes on their way to Richmond Park use exactly the same routes as people pootling to the shops so in many areas, it won't matter if the journey is recreational or utility or if the people are fit or fat, people will want to use the same route.

SpacePootler

The trouble is interpretation, of course. We always argue that you don't justify new bridges by counting how many people swim across.  At what point does "look how many people use this A road!" become "Clearly it's in great shape already. Job done!"

This is where the LCC's "give a beep" campaign showed great promise.  It showed sites of conflict, and allowed us to plot those.  Yes, it highlighted the major A roads, but did so in the context of "this is where the obstacles are".  I'm not sure what policy decisions will come out of this, but it tells an important story that I don't get from strava data even if the routes depicted seem similar.

landscape_urbanism

A local research project could investigate the relationship between alternative data sources:

  • data from a TfL app, of the kind you descripe for West London
  • an on-the-ground census by staff on street corners
  • the Strava app
  • data from the PTC tool (using national census data and  Cyclestreets.net analysis)

It might then be possible to 'translate' one category of data into another eg by applying a correction factor to the Stava data. Could this work?

 

michaelrobinson

TfL do traffic surveys using portable cameras and I believe Automatic Number Plate Recognition allows them to do analysis of origin/destination to understand through traffic versus access traffic for example.

I don't know how these cameras handle bikes.

Also, DfT have average daily vehicle counts with a breakdown of vehicle type from measuring points on A roads.  This is quite high level data but can be useful for some purposes.

Log in or register to post comments