|
|
| Queens Bus Redesign Results In Slower Speeds (350830) | |
 |
|
| Home > BusChat | |
[ Post a New Response | Return to the Index ]
|
|
| (350831) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by New Flyer #857 on Thu Mar 5 08:54:59 2026, in response to Queens Bus Redesign Results In Slower Speeds, posted by Stephen Bauman on Wed Mar 4 22:19:31 2026. Accepting the numbers, can it be explained? I mean if turns and stops are reduced, and presuming the number of traffic signals is similar, what could be holding the buses up? |
|
|
| (350832) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by Stephen Bauman on Thu Mar 5 12:58:01 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by New Flyer #857 on Thu Mar 5 08:54:59 2026. Accepting the numbers, can it be explained?I should have mentioned that this data was calculated from the MTA's published GTFS schedules. It's possible they need a reality check. I was interested in the planning planning results. Somebody at the MTA should have run a similar sanity check and asked "what's going on"? There is another explanation that's a work in progress on my part. It stems from my asking this question: Given that stop distance, turns, traffic signals influence bus speed, what's the relative importance of each parameter? Bus stop spacing is the easiest data to obtain. It's available from every GTFS schedule that includes the optional shape file. Most US operators include this file. Most foreign operators don't. I just finished running a correlation analysis for bus stop spacing and bus speed for 51 of the largest cities in the US. The results showed a rather weak value of 0.27. This suggests that there might be a missing parameter that isn't being considered. I have my own favorite candidate. The correlation came back around -0.76 which means a much heavier negative influence. Further analysis, assuming an exponential or power relation, rather than a linear relation, brought the correlation up to -0.85. NYC's value for this parameter is very high. This means it would swamp out any changes made by increasing stop spacing. I've submitted my analysis to two others for peer review. I'll see what happens. |
|
|
| (350833) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by Lord Vader on Fri Mar 6 08:23:07 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by New Flyer #857 on Thu Mar 5 08:54:59 2026. While the Queens Bus Network redesign affected us in Brooklyn as well, slower speeds can be attributed to many factors. Traffic is number one reason, especially when in my area double parking, sanitation and school buses have caused delays. My primary route, the B47 has had issues since the depot switch. It ran fairly well when it was out of the Flatbush Depot, started getting wonky when moved to Grand Avenue, now operating out of Fresh Pond it’s horrible. Lot of us are concerned about the Brooklyn Network redesign which will affect the B47 coming down our street which many rely on it. I understand the constant turns from Clarkson Ave are a pain but moving the bus to Kings Hwy is a bad move just to put it on the street which causes the most drama, E 98th St. I tried suggesting just leave the B47 on Clarkson and Rockaway Pkwy to ENY Ave and avoid E 98th Street altogether. Truck deliveries for the shops there as well as squeezing through with the B15 are a nightmare.Vader |
|
|
| (350834) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by Stephen Bauman on Fri Mar 6 11:58:20 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by Lord Vader on Fri Mar 6 08:23:07 2026. As I had mentioned to Mr. New Flyer #857, I was considering the only the published schedules, not real time data.The purpose of the redesign was supposed to increase bus speed. The planners' product was to produce new routes and stops that would increase scheduled bus speeds. The planners' product was a new schedule. My point is that the new schedule showed the average bus speed to be less than the previous one. How well any schedule performs in the real world is a valid question. I will be examining that question. My findings will not be able to assess the problems you mentioned because they effect individual trips. I'll be looking at averages. I hope I'll be able to get it down to the route level, possibly by route and hour specificity. |
|
|
| (350835) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by New Flyer #857 on Sun Mar 8 09:47:35 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by Stephen Bauman on Thu Mar 5 12:58:01 2026. I should have mentioned that this data was calculated from the MTA's published GTFS schedules. It's possible they need a reality check.In that case I suspect it's padding. Manufacture increased on-time performance to make the redesign "successful." I personally don't care if they pad the schedules. I just want the buses on time and not bunched (and better on-the-fly actions/solutions when they do bunch). I'd walk further to a bus stop and spend more time on the bus to get rid of bunching. |
|
|
| (350836) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by Stephen Bauman on Mon Mar 9 12:42:32 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by New Flyer #857 on Sun Mar 8 09:47:35 2026. I personally don't care if they pad the schedules.The actual speeds are published with at least a 1 month delay. Here are the actual monthly average speeds for the first 7 months: Jul to Jan before month/year: 7/2024, before avg speed: 8.78, after month/year: 7/2025, after avg speed: 9.02 before month/year: 8/2024, before avg speed: 8.7, after month/year: 8/2025, after avg speed: 9.11 before month/year: 9/2024, before avg speed: 8.49, after month/year: 9/2025, after avg speed: 8.86 before month/year: 10/2024, before avg speed: 8.58, after month/year: 10/2025, after avg speed: 8.93 before month/year: 11/2024, before avg speed: 8.6, after month/year: 11/2025, after avg speed: 9.01 before month/year: 12/2024, before avg speed: 8.62, after month/year: 12/2025, after avg speed: 8.98 before month/year: 1/2025, before avg speed: 9.1, after month/year: 1/2026, after avg speed: 8.9 The January results may or may not reflect the snow storm. We'll have to wait. If there were not a mistake in the MTA's date data, I would have excluded the snow dates. There was some question as to whether these dates were reported to the MTA Board. New bus schedules were issued on 8 Jan 2026, so we'll have to wait and see whether the Jan speed decline is a snow job. BTW, I've communicated the date error to the MTA. We'll see how long it will take them to fix it, if at all. |
|
|
| (350837) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by Italianstallion on Mon Mar 9 16:36:52 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by Stephen Bauman on Mon Mar 9 12:42:32 2026. So these seem to be faster speeds, not slower speeds. What am I missing? |
|
|
| (350838) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by Stephen Bauman on Mon Mar 9 20:33:00 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by New Flyer #857 on Sun Mar 8 09:47:35 2026. The MTA got back to me. What I thought was a bug, turns out to be a "feature". I made a mistake in calculating the speeds in my last post. Here's the the corrected version.before month/year: 7/2024, before avg speed: 8.75, after month/year: 7/2025, after avg speed: 8.96 before month/year: 8/2024, before avg speed: 8.75, after month/year: 8/2025, after avg speed: 9.06 before month/year: 9/2024, before avg speed: 8.53, after month/year: 9/2025, after avg speed: 8.87 before month/year: 10/2024, before avg speed: 8.57, after month/year: 10/2025, after avg speed: 8.91 before month/year: 11/2024, before avg speed: 8.66, after month/year: 11/2025, after avg speed: 9.06 before month/year: 12/2024, before avg speed: 8.69, after month/year: 12/2025, after avg speed: 9.0 before month/year: 1/2025, before avg speed: 8.98, after month/year: 1/2026, after avg speed: 8.92 It would be nice, if the MTA provided raw source data. By that I mean for each trip there would be entries for: trip id that corresponds to the GTFS trip id; arrival/departure timestamps for each stop, stop identification that corresponds to the GTFS stop id; number of passengers on/off the bus at that stop; number of wheelchair on/off at the stop; unique vehicle id for the trip. The don't. Their staff massages the raw data to answer board member questions. This means aggregating the data in various and inconsistent ways. The output is posted on the state's open data website. The result is akin to resolving the elephant in a box problem. |
|
|
| (350839) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by Stephen Bauman on Mon Mar 9 20:36:27 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by Italianstallion on Mon Mar 9 16:36:52 2026. What am I missing?The data isn't current. A comparison with the published schedule for the current time isn't possible. It's the March schedule that showed a decrease in speed compared to the year before. We'll probably have to wait until May to verify the March scheduled speed decrease. |
|
|
| (350849) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by BrooklynBus on Fri Mar 20 13:28:22 2026, in response to Queens Bus Redesign Results In Slower Speeds, posted by Stephen Bauman on Wed Mar 4 22:19:31 2026. I notice that you concluded there are 1,100 fewer bus stops. I calculated almost 1,300.I guess the difference is that if elimination of a bus stops affected two routes, I counted it twice. In your opinion, which would be correct to count it once or twice? |
|
|
| (350850) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by Stephen Bauman on Fri Mar 20 14:59:01 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by BrooklynBus on Fri Mar 20 13:28:22 2026. I described the calculations in my post.I notice that you concluded there are 1,100 fewer bus stops. I calculated almost 1,300. I guess the difference is that if elimination of a bus stops affected two routes, I counted it twice. In your opinion, which would be correct to count it once or twice? You should average over trips not routes. That may account for the difference. In your example, suppose routes A and B operated over a section where 1 stop was eliminated. Suppose further that route A operated every 10 minutes, whereas route B operated every 20 minutes. How many stops were eliminated? |
|
|
| (350854) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by Stephen Bauman on Sat Mar 21 15:41:16 2026, in response to Queens Bus Redesign Results In Slower Speeds, posted by Stephen Bauman on Wed Mar 4 22:19:31 2026. Here are some real time results that is based on the data contained at this site:https://data.ny.gov/Transportation/MTA-Bus-Speeds-Beginning-2015/cudb-vcni/about_data month,before speed,after speed July 2024/2025, 8.68±1.71, 8.77±1.73 August 2024/2025, 8.66±1.70, 8.87±1.74 September 2024/2025, 8.41±1.74, 8.59±1.72 October 2024/2025, 8.48±1.74, 8.65±1.70 November 2024/2025, 8.54±1.77, 8.79±1.77 December 2024/2025, 8.59±1.77, 8.72±1.74 January 2025/2026, 8.74±1.75, 8.90±1.74 February 2025/2026, 8.65±1.73, 8.47±1.76 I've eliminated express buses because I'm interested in local service. The average speeds were calculated by dividing the monthly sums of the distance traveled by the time consumed. This is equivalent to calculating the time weighted mean of the individual route monthly speeds. The standard deviations were calculated as the time weighted standard deviation of the individual route monthly speeds. The numbers differ from a previous post. This is because a problem the MTA's data. They classified some of the QM63 data as local/limited rather than express. I'm also using the MTA's borough classification rather than filtering by Q and not QM as starting characters for the route. I don't know whether trusting the MTA for this filter will also bite me. The salient point is that any difference in the speed is swamped by the standard deviation. The speed differences are within 0.2 mph or only 0.12 sigma units. The best statistical case is that there is no change, despite increasing the distance between stops by 50%. |
|
|
| (350855) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by Italianstallion on Sat Mar 21 16:29:14 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by Stephen Bauman on Sat Mar 21 15:41:16 2026. OK, no change, or slight speed increase; then you should change the heading of your post. |
|
|
| (350856) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by BrooklynBus on Sun Mar 22 15:44:17 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by Stephen Bauman on Fri Mar 20 14:59:01 2026. I don’t see why frequency would affect the number of stops removed.It only matters when considering numbers of affected passengers. |
|
|
| (350857) | |
Re: Queens Bus Redesign Results In Slower Speeds |
|
|
Posted by Stephen Bauman on Mon Mar 23 17:05:05 2026, in response to Re: Queens Bus Redesign Results In Slower Speeds, posted by BrooklynBus on Sun Mar 22 15:44:17 2026. I don’t see why frequency would affect the number of stops removed.It depends upon which question is being asked. 1. How many physical bus stops were removed? 2. How many fewer stops are encountered on bus trips? N.B. I was trying to answer two questions: what's the difference in the distance between bus stops; and what's the difference in average bus speed. One way to answer these questions is to consider all bus trips over a typical period (e.g. 1 week). The GTFS spec provides: total trip distance; total trip time duration; total number of stops. Divide trip distance by number of stops to get the average distance between stops. Divide the trip distance by the trip time duration to get the average speed. You chose to use this data to answer another question. That's fine. The only problem is that one has to be careful about which question the calculation answers. :=) |
|
|