In order to do geographic analysis at a more granular level, I mapped every pickup and drop off to its census tract. NYC is divided into 2,166 census tracts, they vary in size but are typically on the order of 5-10 city blocks wide, with around 4,000 residents each

“Neighborhood tabulation areas” (NTAs) are collections of census tracts that represent distinct neighborhoods of the city. There are 195 NTAs, and they have names you probably recognize: Upper East Side, West Village, Brooklyn Heights-Cobble Hill, and so on

The 5 boroughs are then made up of NTAs. You can view an interactive map of census tract and NTA definitions here, or get static maps here:

This video is processing – it'll appear automatically when it's done.

The taxi data has been around for a little while now, and there are lots of people who have done cool stuff with it. I started working on it in September 2015, but then got distracted for a few weeks, and I’m actually glad I did, because when I got back to it the dataset had expanded from 1.5 years of data to 6.5 years!

Here are some other taxi and Uber-related posts to check out:

As for the maps themselves, one question that comes up a lot: why are there pickups and drop offs on the highways, bridges and tunnels? Certainly people aren’t getting out of cabs on the Triborough Bridge, so what’s up? I’d imagine that it’s mostly hardware error, either the meters submit the data mid-trip, or maybe they just send bad data sometimes. There is a store_and_fwd_flag field in the dataset, which has something to do submitting data after the fact in case the meter did not have a connection at the time of the trip, so maybe those trips tend to be the ones with unreliable coordinates

This video is processing – it'll appear automatically when it's done.

“trying to conserve cash” and “hired David Boies” go together like oil and water. He charges $960 per hour!

This video is processing – it'll appear automatically when it's done.

This is a pretty amazing “summer of 2014” sentence to look back on!

Anyway, this article is great and highly recommended

This video is processing – it'll appear automatically when it's done.

This is almost certainly factually untrue. Probably a more accurate phrasing would have been:

adapting code he’d written to view the Metro-North New Haven Line’s train schedule

But “Yale” plays better in the media, from what I’ve read and heard

This video is processing – it'll appear automatically when it's done.

I didn’t watch this debate (“too well to attend”), but the main narrative seems to be that Bush is dead and Rubio won. At least that’s what gamblers think, since Bush’s betting odds on winning the Republican nomination are down to 9%, while Rubio’s are up to 36%:

In August, Bush has a 44% betting chance, while Rubio was as low as an 8% chance

This video is processing – it'll appear automatically when it's done.