big data needs big tools to sort it

The idea of “big data” is very popular in big business, but it’s trickling down into a lot of other things we use. This should be good; more is better, of course. Well, it is when it doesn’t add to complexity of use. I wanted to share an example of a less than great implementation of big data in a consumer use case…

strava heatmaps

This is the route builder in popular bicycle ride tracking app, Strava. It has a “heat maps” feature that is the big data implementation. It does some mathematical aggregation of all ride traffic to give an idea of how often individual roads are used. The idea, at least I think, is to help you choose “better” bike routes based on the logic that the most people would use the best route. (whatever “best” is a measure of) This is kind of handy in some areas.. Rural places. Small Towns, like Lafayette, where I moved from. But in highly populated areas, like San Diego, where I am now, it’s not as useful. There is a lot of tourist traffic, making some paths, like beachside walkways with high foot traffic, appear to be the right place to ride. My choice of “best” comes from wanting to get to work fast or wanting a hard training ride, so dealing with foot traffic is far from ideal.. but there’s no way to separate it out.

Strava has already started to divide groups by population in other features. “Segments” of routes show a listing by time duration of every user that’s ridden them. This feature has long had a gender segregation, but more recently for paid members offers filtering by age and by weight. Why is this useful? Well, it helps competitive cyclists know how they are doing against other people they might be racing.

I’d really like to see them add to this – it would be great if there was a way to break down the heat map by some categorization. Maybe it’s just max speed on that particular ride. This could weed out the beach cruiser people, or, depending on what you were trying to get out of it, could weed out the folks who are competitive/training.

Regardless, I think this is a decent illustration of lots of data needing more advanced tools to be useful. I think we’re going to be seeing more and more of this as time goes on. It’s a rife place for Interaction Designers to develop new standards.