Everyone knows that New York City is huge, but for transportation purposes the thing that makes it both an exceptionally interesting and exceptionally challenging place for taxis is just how dang dense it can be.  (An aside for you urban planning nerds: if you look only at average density throughout the metro region, you might be surprised to find that LA (gasp!) actually wins out on average.  A more relevant metric called weighted average density confirms what you probably already knew.  Phew.)

If you look at all e-hail requests over a two-week period, you see what you might expect: taxi drivers using Uber get requests in all five boroughs, but demand is concentrated in the financial district and midtown Manhattan (red means more hails).

heatmap

In other words, it’s not just that there’s a lot of people that want taxis – it’s that a lot of people want them at the same place and at the same time.  Now that Uber can be used for e-hailing in New York, we thought it would be fun to see just how our data confirms this observation.

First, I took the above data and aggregated them into “neighborhoods” using an algorithm called k-means (specifically, I used a modified version of a script I found here).  Apologies to all you amateur map-makers out there: I made my best guess naming them by matching the centroids of my polygons to shapefiles provided by the New York Department of City Planning.  The nice folks at the planning department were also kind enough to make available water boundaries so I could clip my shapes to land.

Here are the neighborhoods as you know them (click to see in more detail; don’t complain if I didn’t get your area just right):

neighborhoods

Using a nifty visualization called a cartogram, I can show what each neighborhood would look like scaled by where people are using the Uber TAXI e-hail option:

 

cartogram

 

The second thing to know about demand in New York (or anywhere, really) is that it varies over time.  The city that never sleeps looks more like the city where people all wake up at once: huge spikes in demand occur on weekday mornings, with smaller bumps during the afternoon rush (more on this later).  Interestingly, not having a bar time means that weekend evenings have a shorter peak and longer tail than most places, with the last stragglers heading home around 4 or 5 AM.  (In the image, green = a low number of e-hails, yellow and red show increasing demand density).

New York is also special because of a phenomenon that commuters have felt intuitively but that the New York Times recently quantified. Because of the way cab shifts are staggered, supply goes offline just when demand begins picking up again for the afternoon commute.  Here’s where Uber comes in: our app supports both taxi e-hail as well as Black Car and UberX. Drivers are independent contractors who can drive when they want to and as it turns out, if you let the market work, supply will step up to meet the demand.

Demand