[OC] In NYC, the W is the best line and the B is the worst line if you look at average delays per trip during peak hours

8 comments

1. Source: MTA Delay data

   Tools: Python

2. idc what the data says the NW still sucks

3. Because the W doesn’t go anywhere, nobody needs it.

4. G line doing G line things 

5. Meanwhile, in Japan, trains are so regular that you can practically set a watch by them and they’re issuing full-scale public apologies if they’re leaving 25 seconds early.

   [That’s not a joke, by the way.](https://www.npr.org/sections/thetwo-way/2018/05/17/611860169/japanese-rail-operator-says-sorry-for-inexcusable-departure-25-seconds-off-sched)

6. Which routes carry the most people?

7. anecdotally the Q and the 1 are the most reliable trains. I wonder if there are outliers skewing the means / what this looks with medians instead

8. I think this is one of those weird cases where averages can be quite misleading. I think many folks will use a quadratic loss function to measure delays because, for example, a 5 minute delay feels more than 5 times as annoying as a 1 minute delay.. or put another way, if there is a 2 minute delay on every single train, people get used to it an don’t worry very much, but if that 2 minute average is actually because there is a 20 minute delay on average every tenth train, then it is much more of a concern.
   NYC subways are pretty darned good at getting people from A to B, but regular *long* delays make some lines just feel awful…. Combine that with filthiness that would make a toilet cleaner in a New Delhi slum turn red, and a complete lack of air conditioning in the stations in summer to make the subway seem pretty darned awful to a lot of people.

Comments are closed.