Strava Cycle Dataset for Glasgow - Useful?

Anything that doesn't fit into the boards below.

Moderator: LeeGCU

Forum rules
Don't swear. Be Nice. Superb. More details here.

Strava Cycle Dataset for Glasgow - Useful?

Postby joel_c » 08 May 2014 21:51

I saw this link on Bez's twitter feed (one of my regular twitter lurking haunts)

http://road.cc/content/news/118098-strava-moves-big-data-london-glasgow-already-signed-find-out-where-cyclists-ride

If you read the BTL comments (no, really this time, no bigots) as well as some of the aforementioned twitter chat, the consensus seemed to be that this was a "bad thing". Even if we believe that Glasgow City Council have the staff with the analytical skills to accurately decipher and interpret the data, there remains the problem that the Strava data is hobbled by it being populated by self-selecting users - namely the sort of people who diligently track their mileage and speed on the routes. In other words, committed cycle enthusiasts with a overwhelming sporting bias.

It's therefore likely that, without being placed in the correct context, the people reviewing the data may well have their perceptions and findings skewed by the demographic of users; the kind of people who are comfortable (or at least have adapted to) the hostile conditions out there. Thus the conclusions reached about particular routes and cycle traffic flows might be somewhat warped and not applicable to the kinds of people who find themselves marginalised by the state of our road network. It might be difficult to prove a negative ("what routes would people who don't cycle use?"), but surely other more robust methods (randomised surveys, fieldwork etc), would provide more accurate answers?

But at the same time, as a "data guy", this is kind of my bread and butter; depending on the grain of data, how it can be sliced and diced etc. thus I'm kind of loathe to entirely write it off. Can anyone suggest any positive sides to this story? Is there any worth to the dataset, or are GCC being sold a bit of a turkey?
joel_c
 
Posts: 427
Joined: 26 Nov 2013 17:02

Re: Strava Cycle Dataset for Glasgow - Useful?

Postby hjolreidamann » 08 May 2014 22:21

Are Glasgow City Council buying the data? The Road.cc article, presumably based on a press release, is carefully worded and doesn't mention GCC explicitly, only that Strava "is licensing the results to city authorities and advocacy groups, including in London and Glasgow", and later "customers are located in London, Glasgow". Could be a Glasgow-based advocacy group or private company rather than the council.

Personally I think this data is a good thing. The Strava Metro page notes that in "denser metro areas, nearly one-half of these [Strava activities] are commutes", and it's precisely because different types of rides can be filtered from the data that makes it a so useful. Show me rides between 7am and 10am on weekdays that are reversed in the evening. Show me weekend rides with an average speed below 15kph. That's how the council can use the data well (and exactly what Strava has in mind).

I also think Strava users are more likely to be following desire lines (e.g. Great Western Road) than keeping away from traffic (e.g. following the River Kelvin). A smart council would look for busy cycling roads and choose those for segregated lanes. For example, Strava's heat map of Glasgow shows that as well as using the traffic-free route along the Clyde, cyclists also use the A77, A761, A74, A8, A81, A82, and on and on. Want more people cycling? Put segregated lanes there.
hjolreidamann
 
Posts: 33
Joined: 05 Mar 2014 13:43

Re: Strava Cycle Dataset for Glasgow - Useful?

Postby sallyhinch » 09 May 2014 09:11

Time to Occupy Strava? ;-)
sallyhinch
 
Posts: 147
Joined: 25 Nov 2013 22:04

Re: Strava Cycle Dataset for Glasgow - Useful?

Postby joel_c » 09 May 2014 10:52

hjolreidamann wrote:Are Glasgow City Council buying the data? The Road.cc article, presumably based on a press release, is carefully worded and doesn't mention GCC explicitly, only that Strava "is licensing the results to city authorities and advocacy groups, including in London and Glasgow", and later "customers are located in London, Glasgow". Could be a Glasgow-based advocacy group or private company rather than the council.


I didn't notice that - clearly I didn't read it closely enough :-)

A smart council would look for busy cycling roads and choose those for segregated lanes. For example, Strava's heat map of Glasgow shows that as well as using the traffic-free route along the Clyde, cyclists also use the A77, A761, A74, A8, A81, A82, and on and on. Want more people cycling? Put segregated lanes there.


An equally smart council could conclude that people are happily using these cycle routes and therefore they don't need to do anything! How do you tell which routes cyclists *don't* use because its dangerous versus it being simply unpopular? Will it have sufficient granularity to be able to tell if people are (illegally) cycling on the pavement for their own safety? How does it help identify barriers to cycling? How does it help highlight suppressed demand - the sort of important evidence that could be used to leverage more funding from Government and LAs - if (by it's nature) the data is all about people who are already cycling?

Also, what conclusions can you reach from finding out whether or not cyclists use a given facility or not - if they aren't being used, you might well conclude that your users are skilled and confident enough not to need it. If they *are* being used, you might conclude that they work and are fit for purpose. (Argh!)

As it stands, Strava users will mainly be enthusiasts, which (in my view) might negatively skew the results. I'd maybe give it more credence if people who drive cars were doing this too (indeed, it could be very useful for traffic management if this happened on a large scale - a smart network of signals and message-bards re-directing traffic and adjusting signal timings dynamically based on demand).

But again, I'm really not sure what to think about it - am loathe to dismiss data-driven solutions but you can't discount the quality or context either.
joel_c
 
Posts: 427
Joined: 26 Nov 2013 17:02

Re: Strava Cycle Dataset for Glasgow - Useful?

Postby sallyhinch » 09 May 2014 19:44

I imagine that detours would probably show up fairly clearly in the data - if you know the beginning and end points, and the quickest route, then the amount deviated from that route would be potentially quite interesting
sallyhinch
 
Posts: 147
Joined: 25 Nov 2013 22:04

Re: Strava Cycle Dataset for Glasgow - Useful?

Postby freeyourinnertube » 09 May 2014 20:56

I saw this on road.cc and twitter last night, and was having a meeting today with someone associated with this project

http://futurecity.glasgow.gov.uk/

Which I think is associated with the cycle hack that Joel attended (but I'm not 100% about that).

The person I know is delivering part of the project. I **think** that the strava data may well have been identified as a potential data stream for one part of the project.

My friend is not working on that part of the project, but I have offered some help in testing/user feedback about some of his deliverables via a different group of cyclists that I'm associated with.

I'm not trying to be overly cryptic, but my friend has a few I.P. things to check out before my offer assistance can be accepted, so I hope you understand the sparcity of hard facts.

From my conversation today I emphasised that cyclists are not a homogeneous group and that there are many different types of cyclists and cycling - this wasn't news to my friend. I'm sure I emphasised 8-80 etc. So I hope that reassures a little - though I don't suppose anyone knows what analysis is planned yet.

If you've no data you cannot mine it, but I'm in two minds about this. I understand and appreciate the power of the technology, however it may well be simply another distraction to avoid putting spades in the ground and deploying Dutch specification infrastructure.

A *little* more to go on perhaps.
freeyourinnertube
 
Posts: 68
Joined: 26 Nov 2013 23:44

Re: Strava Cycle Dataset for Glasgow - Useful?

Postby joel_c » 09 May 2014 21:33

Hello,

Going off-topic here somewhat here but ...

Aye its tangentially related - there's an Open Glasgow Future Hack event on at the end of the month. The Cycle Hack organisers mentioned it but said it wasn't a "rival", more something complementary to what they're doing. They characterized the former as being like The Apprentice, where the team work exclusively on one "hack" to get funding at the end of it. By way of contrast, the cycle-hack people are seemingly trying to develop lots of hacks simultaneously and to produce something tangible at the end - i.e. not just a project concept to get further funding.

I understand that the idea is that all the "hacks" at the end are documented and released publicly under either a GPL licence (for the code-y stuff) or a creative commons licence (for original designs, artwork and other IP) - therefore people could use the hacks and develop them further after the event.
joel_c
 
Posts: 427
Joined: 26 Nov 2013 17:02

Re: Strava Cycle Dataset for Glasgow - Useful?

Postby NLCyclist » 17 May 2014 15:37

I reckon the more data like the Strava heat maps, the better. It's really useful for my local area (http://labs.strava.com/heatmap/#13/-3.9 ... /blue/bike) to see where there is demand at the moment.
NLCyclist
 
Posts: 13
Joined: 09 Dec 2013 13:35
Location: North Lanarkshire


Return to General stuff

Who is online

Users browsing this forum: No registered users and 1 guest