[MTAdev] transfers.txt file?

[Dan Coutu]

I note that the Metro North and LI Railroad GTFS files do not contain
a transfers.txt file. Is this file intentionally omitted or
accidentally missing?

Given that the Metro North RR does have transfers on the New Haven
line to the three spur lines I'm guessing that it is missing.
Otherwise how am I supposed to determine the correct transfer stations
and times to jump from main line to spur or vice versa?

Thanks for any insight,

Dan

[Aaron Donovan]

Hi Dan,

You raise a good point. I will bring this up with the railroads.

Yes, there are timed transfers (guaranteed connections) on the New
Haven Line at Stamford for the New Canaan Branch, South Norwalk for
the Danbury Branch and Bridgeport for the Waterbury Branch, but this
is by no means an exhaustive list of the Metro-North/LIRR transfer
points.

-Aaron Donovan
Metropolitan Transportation Authority

[John L]

MNR omits this intentionally.

the current transfers.txt cannot differentiate a connection between lines. for instance GCT to New Canaan would connect trains at 125th street erroneously as there are no connections for this trip at that location. it would also try to connect at Greenwich but this would be for local service westbound only. a real connection only exists at Stamford.

I can share a proposed transfers.txt file that contains route from, route to, stop from, stop to, transfer type, min transfer time, max transfer time. pulling this from memory so bear with me if a field is missing.

let me know.

regards,

John

[Dan Coutu]

John, please do share what you have. If you need assistance from the group with refining it further to make it more useful then please say so.

It does seem to me that if the original transfers.txt file can't differentiate connections between lines then either it has the wrong/insufficient data or the GTFS is inadequate regarding transfers.

Thanks!

Dan

[John L]

I think this is the fastest way to post this. I think that MAX_TRANSFER_TYPE should be included as 1200. The rule MNR uses is between 6 and 20 minutes to make transfers for valid trips. Don't want people waiting on platforms for no reason. There are cases when MNR makes connections less than 6 minutes but those are for connections that will wait. Trying to get this as a transfer_exception.txt. There are about 1% of trains that make this type of connection. So by no means is this a silver bullet.

Transfers are trips that can happen, Connections are trips that are scheduled to happen.

STOP_FROM,STOP_TO,ROUTE_FROM,ROUTE_TO,TRANSFER_TYPE,MIN_TRANSFER_TIME

20,20,1,1,2,360

27,27,1,1,2,360

33,33,1,1,2,360

39,39,1,1,2,360

1,1,1,2,2,360

4,4,1,2,2,360

1,1,1,3,2,360

4,4,1,3,2,360

1,1,1,4,2,360

4,4,1,4,2,360

1,1,1,5,2,360

4,4,1,5,2,360

1,1,1,6,2,360

4,4,1,6,2,360

1,1,2,1,2,360

4,4,2,1,2,360

62,62,2,2,2,360

68,68,2,2,2,360

74,74,2,2,2,360

83,83,2,2,2,360

94,94,2,2,2,360

1,1,2,3,2,360

4,4,2,3,2,360

56,56,2,3,2,360

1,1,2,4,2,360

4,4,2,4,2,360

56,56,2,4,2,360

1,1,2,5,2,360

4,4,2,5,2,360

56,56,2,5,2,360

1,1,2,6,2,360

4,4,2,6,2,360

56,56,2,6,2,360

1,1,3,1,2,360

4,4,3,1,2,360

1,1,3,2,2,360

4,4,3,2,2,360

108,108,3,3,2,360

112,112,3,3,2,360

116,116,3,3,2,360

124,124,3,3,2,360

138,138,3,3,2,360

155,155,3,3,2,360

124,124,3,4,2,360

124,124,3,5,2,360

131,131,3,5,2,360

124,124,3,6,2,360

140,140,3,6,2,360

124,124,6,4,2,360

Hope this helps!

John

[David Turner]

When a station is split into eastbound and westbound sides, and there is
no good connection between the sides, it should be split the stop into
two stops in GTFS.

This is actually a suggestion for the subways too. Here is an example
of a trip that is wrong because it requires that a transfer be made from
a downtown train to an uptown train at a stop where that transfer is
impossible:

http://maps.google.com/maps?f=d&source=s_d&saddr=jfk&daddr=101st+st+at
+manhattan+ave&hl=en&geocode=FbgWbAIdlv2Z-yn1TphITWbCiTE_zG0Zk0STuA%
3B&mra=ls&dirflg=r&ttype=dep&date=05%2F21%
2F10&time=10:51am&noexp=0&noal=0&sort=&sll=40.795569,-73.961549&sspn=0.01168,0.017767&ie=UTF8&ll=40.765136,-73.98335&spn=0.011685,0.017767&z=16&start=0

This will become even more important when the MTA starts putting
wheelchair accessibility information into their GTFS.

[Sklar, Robert]

The link in your message describes a trip from New York to Wichita, KS!

[Kaufman, Sarah]

Thanks, David, for bringing up the subways transfer discussion, because I'd like to revisit this topic as well. A few months ago I put out a request for a helping hand on developing a great transfers.txt file for the subways. I didn't get any responses, but I'd like to re-submit the request, now that more people are on the list (and we still don't have funding to replace the person at NYCT responsible for generating this file). If anyone reading this interested in working on the transfers file for subways, please contact me at: sarah....@nyct.com

No need to plan trips from here to Wichita, just within the NYCT property. :)

Thank you,
Sarah

-----Original Message-----
From: mtadevelop...@googlegroups.com [mailto:mtadevelop...@googlegroups.com] On Behalf Of David Turner
Sent: Friday, May 21, 2010 11:30 AM
To: mtadevelop...@googlegroups.com
Subject: Re: [MTAdev] transfers.txt file?

[Michael B. Justice]

Sarah:

I'd be interested in helping.


MJ

[David Turner]

Sure, if you just copy the first line. Try this link if your mail
reader is broken:

http://bit.ly/9gRjft

[David Turner]

A transfers file is insufficient, because the two sides of 50th St are
still one stop. Also, I think my colleague Nick sent in a proposed
transfers.txt some time ago:

http://github.com/novalis/ImproveMtaGtfs/blob/master/subway/transfers.txt

Note that this link is probably broken right now, because Github is
having some issues; you can still check it out by running:

git clone g...@github.com:novalis/ImproveMtaGtfs.git

(or just wait a bit)

[John Larsen]

David,

If the transfers.txt looked like the following

STOP_FROM, STOP_TO, ROUTE_FROM, ROUTE_TO, TRANSFER_TYPE,
MIN_TRANSFER_TIME

And using your trip above the transfer would look something like
50th, 50th, E, C, 0, 0 // in reality I think it should be 50th, 50th,
E, C, 2, 420 to give some time for the non sprinters, as i think you
have to hike to get to the other platform

Wouldn't that resolve the problem?

Granted, this does not resolve ADA issues. I think adding data to the
stop itself and maybe adding a fourth transfer_type or an additional
column for ADA transfer with a min_transfer time so that they can be
differentiated. Using this fourth type or added column if an ADA trip
was requested and only if the stop is full ADA for transfers or
partial ADA if it is an origin or destination. This is another
discussion altogether.

On May 21, 12:21 pm, David Turner <nova...@novalis.org> wrote:
> A transfers file is insufficient, because the two sides of 50th St are
> still one stop.  Also, I think my colleague Nick sent in a proposed
> transfers.txt some time ago:
>

> http://github.com/novalis/ImproveMtaGtfs/blob/master/subway/transfers...

>
> Note that this link is probably broken right now, because Github is
> having some issues; you can still check it out by running:
>
> git clone g...@github.com:novalis/ImproveMtaGtfs.git
>
> (or just wait a bit)
>
>
>
> On Fri, 2010-05-21 at 11:56 -0400, Kaufman, Sarah wrote:

> > Thanks, David, for bringing up the subways transfer discussion, because I'd like to revisit this topic as well. A few months ago I put out a request for a helping hand on developing a great transfers.txt file for the subways. I didn't get any responses, but I'd like to re-submit the request, now that more people are on the list (and we still don't have funding to replace the person at NYCT responsible for generating this file). If anyone reading this interested in working on the transfers file for subways, please contact me at: sarah.kauf...@nyct.com

[David Turner]

On Fri, 2010-05-21 at 14:33 -0700, John Larsen wrote:
> David,
>
> If the transfers.txt looked like the following
>
> STOP_FROM, STOP_TO, ROUTE_FROM, ROUTE_TO, TRANSFER_TYPE,
> MIN_TRANSFER_TIME
>
> And using your trip above the transfer would look something like
> 50th, 50th, E, C, 0, 0 // in reality I think it should be 50th, 50th,
> E, C, 2, 420 to give some time for the non sprinters, as i think you
> have to hike to get to the other platform
>
> Wouldn't that resolve the problem?

No, because there is no transfer available there at all. There's no
overpass or underpass. The correct route is to transfer at 42nd street.
If you have an unlimited metrocard, you could go above ground at 50th,
but it is almost certainly wrong to give these trips.

> Granted, this does not resolve ADA issues. I think adding data to the
> stop itself and maybe adding a fourth transfer_type or an additional
> column for ADA transfer with a min_transfer time so that they can be
> differentiated. Using this fourth type or added column if an ADA trip
> was requested and only if the stop is full ADA for transfers or
> partial ADA if it is an origin or destination. This is another
> discussion altogether.

ADA issues are way more complicated than this. I am working on ADA
issues in GTFS separately offline, and we hope to have a proposed
standard soon.