How does Strava's heatmap fair in predicting nicer routes?

http://labs.strava.com/heatmap/

This is also partly a reply to cardamomo's post. I imagine that tracking the routes commuters actually take is a great way to find the better routes.

That's my go-to reference when looking for new routes. We can really think of its reliability as a signal-to-noise ratio: inexperienced road cyclists might cause the map to show significant traffic on bad roads while experienced cyclists choose good roads. There are presumably fewer cyclists entering the realm of road cycling than there are experienced road cyclists. The new road cyclists are likely aware of the dangers of the sport, so in most cases, they probably seek resources for planning rides (e.g., cycling groups).

That brings up another point: Group rides will generally use good roads, and there will be multiple cyclists simultaneously contributing to the same route on the heat map. The boldest paths on the heat map may then very possibly be routes ridden by group rides, suggesting they're very safe.

More experienced cyclists might be more likely to own and use GPS units as well. Strava's heatmap only has data from users that self-report GPS traces.

Strava has apps for iPhone and Android so the barrier of entry is lower than owning a dedicated GPS unit: https://support.strava.com/hc/en-us/articles/223297187-How-t...

Oh, you're right. I totally forgot about that.

Strava's heatmap is great. I use it all the time.

> Part of the trouble is that street markings do not reflect car traffic or other special circumstances.

I absolutely agree. I bike almost daily in Brooklyn, and the information I glean from maps like this (or the official NYC bike map) is only a small part of what I consider when selecting a route. Having experience with the nature of car and truck traffic in a given neighborhood is an order of magnitude more valuable than information about bike lanes.

Cyclist safety is a tough issue. From an infrastructure perspective, creating safe bike lanes is a hell of a lot more difficult than simply painting lines on a street. I wonder what ways tech can help cyclists make safe choices beyond maps.

I had to RTFM to figure out what the yellow highlights are. Those are roads that have been designated as part of a biking road network. In your example Fremont should presumably be highlighted yellow. I think the renderer puts far too little visual emphasis on the bike network roads (and omits the markings from the key!)

Since it is open street maps, it relies on manual curation. My guess is that some curators use the official city bike maps, and others editorialize.

This seems to do a great job where I am, though it crashes repeatedly on both iOS devices I tried. It definitely needs better testing with mobile safari (and an app with some sort of tile caching / offline mode, unless I'm missing something)

> I had to RTFM to figure out what the yellow highlights are.

Yeah, I was confused at first too. It is mentioned lower down in the article.

> In your example Fremont should presumably be highlighted yellow. I think the renderer puts far too little visual emphasis on the bike network roads (and omits the markings from the key!)

Yeah, although the Seattle yellow highlights are missing a lot (e.g. the Westlake cycletrack, which is one of the few "green" corridors in the city and used by tons of cycle commuters, isn't highlighted). The Elliot Bay trail, another "green" corridor along the bay, is highlighted around the interbay area but not the portion that heads into downtown. The Interurban North multi-use path is also missing from the highlight. But that's just bad data; it can be fixed.

Edit: Oh wow. It turns out some highlighting isn't shown at some zoom levels. (In particular, the zoomed out view I would use to try and plan a route is missing some highlighted segments that show up at higher zoom.) I saw a random block highlighted yellow; zoomed in and it is part of a much longer segment. That is a big UI problem.

(Fremont Ave is highlighted at that higher zoom.)

The preferred styling is OpenStreetMap. One word, no sss.

Part of what loeg is saying is that the dedicated bike infrastructure they have is not especially bike friendly. To the extent people are following OpenStreetMap's "on the ground" rule, it will be hard to extract information about that from OSM.

Here the main highways are US Bike Routes, which means they get highlighted on the Mapzen style, with the exception of someone doing a through trip (where they are the logical route), they are the absolute worst places to cycle in town.

So OSM needs tags for doorzones?

Yeah, until cities stop creating such bike lanes :-).

It does have a width tag, and it should be applicable to lanes ;) That's a bit of a stretch, I admit.