Mapping San Francisco, New York, and the World with Uber
It’s been an Uber busy couple of weeks. We officially launched in New York, gave some not-so-subtle hints at our upcoming expansion, and made a big push to get Travis’s fly wardrobe prepped! To top it all off we’ve moved into a shiny new office. Last time on the #UberData channel we talked about the hidden costs of taking a cab by exposing how inefficient cab dispatch is and how long you’ll have to wait for a cab if they even decide to show up. That post didn’t even take into account how much nicer our cars are and the fact that San Francisco is apparently considering increasing cab fares for the first time since 2003.
In this post we’re going to get a little more down and dirty with the details of how Uber’s Team Science are working to make sure you get an Uber as quickly as possible. To do this we mainly need to account for 4 things:
1.) Where are people coming from?
2.) Where are people going to?
3.) Where are our drivers at?
4.) When do these patterns change… during?
The nice thing about having a data-driven company is that we can actually learn from our users by applying science. Put simply: when we know where our drivers are when they accept a fare, we can assess inefficiencies in our pick-ups by comparing our drivers’ locations to where are riders are when they want to be picked up.
This is a map of where our drivers were when they accepted a ride. The more white the color, the more likely it was that our driver was at that exact spot when they accepted the ride. Look at our coverage! What’s neat is how certain features of SF pop out, such as the Golden Gate and Bay bridges and the 80, the 280, and the 101 (yeah, I said THE… I grew up in San Diego). Mission and Market also pop out, as do the Embarcadero and Columbus on the east side, and Balboa and Geary in the Richmond. You can also see a couple of other highlights like the circular layout of SFO or the de Young and California Academy of Sciences in Golden Gate Park.
Remember in the last #UberData post, the math arrow everyone seemed to love?
In that graph we showed how we, as a company, are getting better every month at reducing our pick-up times for our riders.
As Travis said in our Wired article, a lot of those improvements were due to better demand prediction: if we know we’re going to have more riders on Saturday night than Tuesday morning then we’re going to need more cars around the city waiting for you on Saturday. But that will only get us so far.
We’ve also been working to optimize pick-up times by incorporating our knowledge of where people are, where they want to go, and when those factors change. In the TechCrunch article covering our NYC launch, we shared a public glimpse into some of our Uber data and ops management. For example, we gave you a peek at “God View”:
which is the name for our real-time rider mapping system with which we can examine how our system is flowing. This is an important complement to our data analyses because it allows our ops team to make rapid changes based on real-time data to correct for deviations from our mathematical predictions.
On top of that, our newest data scientist, Henry Lin, has been rocking the demand prediction stuff. In that same TechCrunch article we showed Henry’s map of expected wait times in SF:
From that map you can see that if you’re anywhere east of Golden Gate Park and north of about Noe Valley, you’ve got an average expected wait time of less than about 5 minutes, easy. If you’re in the Richmond or Sunset the wait times hit about 10-12 minutes.
These times are crazy fast compared to cabs. But we can do better. With science!
Again, click on that map to zoom in. We’ve blacked out all of the details of the city to highlight the analytics. This is a map comparing where people in San Francisco are going to versus where they’re coming from. The more red the map is, the more people go there. The more blue, the more people come from there. From this map it’s clear:
Uber riders want to go downtown, Hayes Valley, the Mission, Van Ness, OAK and SFO. They’re coming from Soma, Pac Heights, and the Marina.
Business in the front, party in the back! But that’s not all. We can also map general interest in Uber. Trust me when I say we take this into account when we’re working out which cities to expand into next.
This is a map plotting the number of times someone has looked at our mobile app, where the more white the color, the more views we’ve gotten. You can see SF (obviously), LA, and San Diego in California. Rounding out the west coast are Portland, Seattle, and Vancouver. A little farther east we’ve got Phoenix, Vegas, and Denver. Centrally there’s Austin, Dallas, Houston, New Orleans, and Dearing/Coffeyville, Kansas (“Center” of the US in Google Maps). In the midwest, Chicago really stands out along with Minneapolis. In the South, Atlanta, Orlando, and the southern Florida coast from Miami up to West Palm Beach are bright. Finally, in the northeast, you can see a nice line from DC to Boston covering Philadelphia and New York, as well as Toronto.
By zooming in on a city and looking at who’s been checking us out, we can build a prediction of where our cars need to be before we even get there.
In Manhattan there are bright hotspots of interest in Flatiron district, Soho, East Village, West Village, and across the bridge in Williamsburg, with some density in midtown. These kinds of maps inform our team where we need to start keeping our cars. So if you’re a fan of Uber, and you want us in your city, let us know! Sign up, open our app, and we’ll take it from there.
Today, we are excited to announce that Uber will give $5.5M to support a new robotics faculty chair as well as three fellowships at CMU. This gift is part of a partnership we announced earlier this year. In addition, we’re pumped to be part of a growing innovation ecosystem in Pittsburgh that includes world leading research institutions and companies, as well as an increasing number of start-ups.
A new report conducted in partnership with Mothers Against Drunk Driving (MADD) reveals that when empowered with more transportation options like Uber, people are making better choices that save lives.