NextBus is least accurate during commute time, study says
NextBus is least accurate during peak commute time, with the 82X, 28, Muni Metro Bus Shuttle, 81X, and 39 routes ranking the lowest in prediction accuracy, according to a new study by Swyft, a mobile transit app. The study looked at NextBus prediction data from August 2015, comparing it to actual arrival times, and defines accuracy as “if the actual arrival time of the vehicle is anywhere between 30 seconds earlier and 4 minutes later than the predicted arrival time.” The study found that the most accurate routes were the 6-Haight/Parnassus, 35-Eureka, the 88 BART Shuttle.
NextBus uses GPS data on individual buses to predict their arrival time, similar to Google Maps and other services. The new study found that when a vehicle is five minutes away from its stop, NextBus is accurate 91 percent of the time, but when the bus is about 25-30 minutes away from its stop, NextBus accuracy drops to 59 percent.
How workable this is for you as a daily rider depends on how much you care about when you arrive at the bus stop. In the latest Muni rider annual survey, riders listed “more frequent service” as their top request, with 11 percent listing “on-time service” as their top issue. After all, if you’re at the bus stop, it’s probably more important that the bus is coming in the next couple of minutes, rather than knowing that the bus is really accurately 77 minutes away (see image above).
Kim Gregory from NextBus told Muni Diaries that “In general, our agency customers are focused on the 0 to 5 minute window and the 0 to 10 minute window.” Gregory says that “during peak periods in urban environments the time between buses may only be 8 to 20 minutes, so people aren’t looking out 30 minutes for a bus. They want to see the next one or two.”
Swyft’s CEO Jonny Simkin says he wants riders to submit real-time data about transit issues using his app, and believes that it is “possible to improve the algorithm by having it factor in these real-time issues.” Currently, riders can use the app to report delays and other issues at their stop, which will create an alert for other riders using the app. However, though you can currently report a delay issue at your Muni stop via the app, the data currently does not change the NextBus prediction time for riders at another stop farther down the route. Simkin says he hopes to share the data with SFMTA to improve prediction accuracy.
Crowdsourced data could certainly help transit agencies, like how Waze shares its driver-reported data with Los Angeles County and city. Swyft’s study says that while their current data “does not solve prediction accuracy problems, it demonstrates the potential for a broader community to work together to avoid common transit issues and delays.”
Photo by Andy B.
The 71 was CURSED – sometimes the arrival would drift and drift and then just disappear.
I wonder if they included NextBus’s notoriously inaccurate turnback times for when buses and trains reach the end of the line and wait a seemingly random interval before going back into service?
Thought 27 bus was unpredictable. It comes and goes as it pleases.
All too often I’ve had to miss several buses because of the blatantly wrong prediction times. I step away from the bus stop to get something to snack on, and then the bus just passes by non-chalantly. It becomes like that SpongeBob SquarePants episode “Rock Bottom”
omg spongebob bus is here! 😀
As I heard at a meeting from someone who intimately knows NextBus: shit in, shit out.
This is definitely one of my biggest Muni problems. It is ridiculously hard to plan for any kind of timely arrival anywhere without simply allowing for a 30-45 minute cushion. Drivers routinely depart from stops well before their scheduled time, or sometimes a scheduled departure never leaves at all. The express buses downtown from the Richmond are especially bad in this regard. I’ve given up altogether on taking the N or T from the Caltrain stop in the evenings.
The 29 is rarely on time during commute hours and often doesn’t show up at all. A couple of weeks ago the driver stopped the bus to go shopping at a corner store and left the bus running with everyone inside waiting. Strange.
I appreciate Swyft effort to highlight NextBus accuracy challenges. Real time data input from riders could significantly enhance prediction algorithms potentially improving the overall transit experience.
Solar Panel Cleaning Services in Chino Hills