How to estimate or obtain the number of trips between each stops?

[KejiDartmouth]

Hi Folks

I am doing a research which using the NYC public transit as the case study. Now the bottleneck of my current model input is to correctly estimate passenger behavior on weekday peak hour. So my question is that is there any public data to display number of  stop to stop trips at peak hours? If not, is there any way to approximate those numbers?

Thanks

Keji

[bobtehpanda]

2 hours ago, KejiDartmouth said:

Hi Folks

I am doing a research which using the NYC public transit as the case study. Now the bottleneck of my current model input is to correctly estimate passenger behavior on weekday peak hour. So my question is that is there any public data to display number of  stop to stop trips at peak hours? If not, is there any way to approximate those numbers?

Thanks

Keji

Because there is no requirement to swipe/tap out, this data doesn't really exist.

[CenSin]

I’ll tell you a few things that apply to most passengers:

1. Cross-platform transfers wherever possible. Most passengers do not like stairs. If there is another station that can get them the transfer with less physical effort, they will take that opportunity instead. If there is another non-cross-platform transfer that takes less effort and does not incur additional travel, passengers will flock there instead.
2. Express trains versus seats: I feel they are about even. Some passengers will not give up seats for a faster ride. Others will transfer and stand if the ride may be faster.
3. Buses: passengers will not take one unless the subway does not go there and/or the subway route is so circuitous that the bus is faster.
4. Not all subway stations are accessible. The bus may be the only way around if not car service.

[RR503]

The MTA has an internal O/D model that they create using surveys and regional data sets, but I don’t think they’d be willing to share. 

What’s the exact purpose of your research? If the accuracy of transfer distributions doesn’t matter too much, you could, in a pinch, use CTPP and then guesstimate routes using elasticities?  

[KejiDartmouth]

16 minutes ago, RR503 said:

The MTA has an internal O/D model that they create using surveys and regional data sets, but I don’t think they’d be willing to share. 

What’s the exact purpose of your research? If the accuracy of transfer distributions doesn’t matter too much, you could, in a pinch, use CTPP and then guesstimate routes using elasticities?  

Thanks for your patient reply. What is CTPP? I am the rookie of this forum.

The purpose of my research is to understand how many passengers are traveled within each borough by public transit.

Thanks

Keji

[RR503]

15 minutes ago, KejiDartmouth said:

Thanks for your patient reply. What is CTPP? I am the rookie of this forum.

The purpose of my research is to understand how many passengers are traveled within each borough by public transit.

Thanks

Keji

CTPP is Census Transportation Planning Products — it’s a really detailed dataset that gets down to the census tract level (maybe even the block group level; it’s been a second since I’ve used) and allows you to view commutation between two tracts, the mode of that commutation, etc. Flaw is of course that not all trips are commutation and that not all commutation trips follow perfect shortest path from O to D, but it should get you started. 

If you’re just looking for figures for transit trips by borough, though, I’d look at bus/subway ridership stats. You’d honestly get a more accurate picture that way, and with the bus redesign effort forcing the agency to publicize some more granular bus network data, you’d be able to have some specificity too. 

[KejiDartmouth]

5 hours ago, RR503 said:

CTPP is Census Transportation Planning Products — it’s a really detailed dataset that gets down to the census tract level (maybe even the block group level; it’s been a second since I’ve used) and allows you to view commutation between two tracts, the mode of that commutation, etc. Flaw is of course that not all trips are commutation and that not all commutation trips follow perfect shortest path from O to D, but it should get you started. 

If you’re just looking for figures for transit trips by borough, though, I’d look at bus/subway ridership stats. You’d honestly get a more accurate picture that way, and with the bus redesign effort forcing the agency to publicize some more granular bus network data, you’d be able to have some specificity too. 

THANKS.

I agree it is better to ook at bus/subway ridership stats. But tbh, I haven't any public  data which listed the information as I want to obtain.

Would you mind sharing me the link if possible?

 

 

[RR503]

29 minutes ago, KejiDartmouth said:

THANKS.

 I agree it is better to ook at bus/subway ridership stats. But tbh, I haven't any public  data which listed the information as I want to obtain.

 Would you mind sharing me the link if possible?

Here you go!

CTPP:

https://ctpp.transportation.org/2012-2016-5-year-ctpp/

LEHD (another, less intensive but more accessible dataset that measures commute dynamics) 

https://onthemap.ces.census.gov/

NYCT ridership figures: 

http://web.mta.info/nyct/facts/ridership/

Bronx bus redesign existing conditions report, which has some good information on intra-boro ridership patterns:

https://new.mta.info/sites/default/files/2019-02/Bronx Bus Network Redesign Existing_FINAL.pdf

1940-1995 subway ridership by station (yours truly put this together):

https://drive.google.com/open?id=1g3pNbvpfqGE4g9v-7t3bfU0xO23U40fp

1998-2016 subway ridership by station (by Nicole Gelinas via Twitter):

https://docs.google.com/spreadsheets/d/1-_GLA3fT0fLgfjX2JLcuKYYJQCxc7HDV5T86WMJTFpI/edit?usp=sharing

...and if you're interested, NYMTC tracks hub bound people flows granularly going all the way back to 1973:

https://www.nymtc.org/Data-and-Modeling/Transportation-Data-and-Statistics/Publications/Hub-Bound-Travel

Edited May 25, 2019 by RR503

[bulk88]

MTA does O/D by PM entry swipe, and assigns PM entry swipe as exit location of AM swipe. Assigning express vs local, and which transfer is used is another whole set of controversy. IIRC N is 2 minutes faster to Time Sq than 7 from Queensboro Plaza, but you always take 1st that shows up, since there are 2-3 7s for every N and the sheer frequency of the 7, the 7 will win, do not let a 7 pass waiting for the N. Also 6th ave. B/D DO NOT beat a parallel local F/M between 34 and West 4, or at best 15 seconds faster. 

Edited May 29, 2019 by bulk88

[bwwnyc123]

Just like on 8th Ave between 59th St and Chambers St/World Trade Center. If the local come first and no express take local.

[RR503]

2 hours ago, bulk88 said:

MTA does O/D by PM entry swipe, and assigns PM entry swipe as exit location of AM swipe. Assigning express vs local, and which transfer is used is another whole set of controversy. IIRC N is 2 minutes faster to Time Sq than 7 from Queensboro Plaza, but you always take 1st that shows up, since there are 2-3 7s for every N and the sheer frequency of the 7, the 7 will win, do not let a 7 pass waiting for the N. Also 6th ave. B/D DO NOT beat a parallel local F/M between 34 and West 4, or at best 15 seconds faster. 

Yeah, express travel and any non perfect commute trip is where the model breaks down. 

Also, FWIW, median B/D runtime is about 1-1:30 less than FM in both directions...not enough to tip the scales, sure, but worth noting nonetheless. 

[Caelestor]

On 5/29/2019 at 9:29 AM, RR503 said:

Yeah, express travel and any non perfect commute trip is where the model breaks down. 

Also, FWIW, median B/D runtime is about 1-1:30 less than FM in both directions...not enough to tip the scales, sure, but worth noting nonetheless. 

Seems about right, based on schedules a skipped stop saves about 45 seconds. Express service only really makes a big difference on the IRT due to its really closely-spaced stops, the BMT Brighton line, and the long express stretches on the IND (CPW, QBL, Fulton). Otherwise, treat it as another subway line with fewer stops to save construction costs and riders' time.