Thursday, August 14, 2014

Average vehicle speed in NYC

Ben Wellington at I Quant NY has a post up about trying to identify the beginning and end of NYC's "rush hour" using the Taxi and Limousine Commission's dataset of 2013 taxi trips.

(Let me echo here his call for TLC to properly open their data - they responded well to Chris Whong's Freedom of Information Law request, but we should not be made to beg for access to data accumulated by a public agency supported by taxpayer money.)

Ben's basic findings are about what you'd expect: most New York City car trips are rather slow. The average speed of all taxi trips was only 13.3 miles per hour; between 8 am and 7 pm that number is just 11.5 mph.

Credit: I Quant NY
I have some hesitations about Ben's analysis. He correctly notes an inherent flaw in his dataset - namely, taxi trips are not a representative sample of all vehicle trips; they are heavily skewed toward the central business district, where congestion is greater and vehicle speeds are lower.

Beyond that, however, I would have liked to see some clarification about whether that 13.3 mph overall average speed was based on the crow-flies distance between each trip's start and end points, or took into account approximations of each trip's street route, the way Chris used the Google Directions API for his mesmerizing visualization

Absent any explicit statement, I have to assume the speed was based on crow-flies distances. (If Ben, or Chris, had run all 172 million trips through the Google Directions API, I would expect to have heard about it!) I am going to very roughly estimate the sinuosity of the average trip as about 1.3, which would mean the 13.3 mph figure would translate into something more like 17.3 mph average speed "as the taxi drives," which feels more accurate considering all those late night free-for-alls and long highway trips to and from the airports.

Ben's speed graph also presents a good opportunity to discuss the dynamics of NYC taxi schedules. He notes that speeds are highest at 5:18 am, but suggests that this is because "the least traffic [is] on the road." Rather, the local peaks in average trip speed around 5 am and 5 pm can be attributed to the taxi driver shift changes that commonly occur around those times, when drivers return to fleet home bases often located in Queens.

After those shift changes, drivers frequently seek passengers at JFK and LaGuardia airports, which leads to long trips mostly on highways. (Even congested highways generally flow faster than surface roads, at least in New York.) This dynamic is a major reason you can never get a cab during the afternoon rush, and, as I think about it, the codification of this system, which meets drivers' needs but not passengers', is probably helping Uber and now also Lyft demolish the taxi industry.

On the whole, though, the I Quant NY post does accurately reflect a fundamental truth of NYC transportation: if you want to get there on time, skip the car! I can easily exceed the taxis' adjusted average daytime speed of 15 mph on my bicycle, and the subways average 17.4 mph night and day, rain or shine. "Go ahead and skip that cab to lunch," says Ben; go ahead and skip most other cab rides, too, say I.

(You can use the time you save to lobby for increased capital funding for the MTA.)