From Uber’s website:

These numbers are UberX only, so the more expensive UberBlack fares will make the averages higher

Why are fares decreasing? Almost certainly part of it is that UberX cut its rates in July 2014, but there might be other factors too. Maybe when Uber first came to NYC it was more popular with corporate clients, who were more likely to take longer trips, then as the service became more ubiquitous people started to use it for shorter day-to-day trips

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

From the report:

Say that the 42% who work 1–15 hours average 2 days per week, the 35% who work 16–34 hours average 4 days per week, and the remaining 23% average 6 days per week. That averages out to 3.6 days per week

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

Worth noting that Jonathan Hall is an Uber employee, and Professor Kreuger “acknowledges working on
this report under contract with Uber. The terms of the agreement granted Krueger ‘full discretion over the content of the report.’”

I have no reason to doubt the veracity of anything in the report, but it was commissioned and paid for by Uber, so it’s okay to be a bit suspicious. If nothing else, had the researchers discovered something that looked bad for Uber, they probably wouldn’t have published their findings

https://s3.amazonaws.com/uber-static/comms/PDF/Uber_Driver-Partners_Hall_Kreuger_2015.pdf

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

I’m not sure if the same car is allowed to drive for both Uber and Lyft. If so, those vehicles might be double counted. I didn’t see anything obvious about this online, but I feel like I’ve been in Uber cars that had Lyft apps too, though I’m not 100% sure about that…

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

Some of the vehicle data seemed obviously wrong, e.g. for a single base there were 1000 vehicles for 5 consecutive weeks, then 2000 vehicles in one week, then 1000 vehicles for another 5 weeks

I removed these suspicious rows from the graph, here’s the relevant part of the code that flags whether the vehicle data appears unreliable

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

A for-hire vehicle (FHV) is the TLC’s term a car that provides prearranged service. Every FHV is associated with a TLC-licensed base, and I used the bases' “doing business as” labels to determine which bases were Uber or Lyft

See here for the section of the code that identifies which bases are Uber or Lyft; it’s a very simple string check

There are 29 different Uber bases in the data, but only 1 Lyft base (doing business as “LYFT-TRI-CITY”)

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