Timeline Analysis of the OC Transpo GPS Data

[Shawn Hooper]

Hello All,

It was great meeting those of you made it to the brainstorming/FAQ session last night.

I've been thinking about the development of the data capture project Edward suggested. 

The basic idea is to capture results from calls to the API over time to do trend analysis.  Given that we need to call each stop individually to get results, I could see this involving a large number of calls (each stop called once an hour, for example) to the API.  It would be easier if we could call the API for a route number.  It would lessen the number of calls required.

I know we're going to be getting access to new parts of the API, is one of them a route # query?

Also, Edward, you said that there was a professor at (I believe) U of O who was interested in this?  Do you have his/her contact information? Are they on this list?

Cheers.

Shawn

[Darren Kipp]

I was thinking of doing some things that would involve some collection
too. It might be worth while to to get together on that prior to the
API such that we are not repeating calls upstream.

I too would like to be able to perhaps query buses (say by run number)
or routes. This would greatly limit the number of times we are
hitting the API since most processing will be done locally.

I could see applications where a predictor could be built based on the
performance of prior buses along a particular route segment.

Another thing that would be really nice is a run id, so things could
be more easily mapped back to the static schedule data.

[Daniel Langer]

If there's a group consensus that having a large chunk of historical data would be a good thing, what about collecting it at the (proxy) server level? If Jonathan's got the time to do so, what about having (for example) the wrapper copy the results of all calls that all users make, recording them and the results (with no information about who made the calls), and making that available as a daily data dump or something similar?

[Jonathan Rudenberg]

Before we take on this project I think it's worth contacting Glen
McGregor. This is his post from the old DataOtt.org site this summer:

Posting by Glen McGregor on Jul 21, 2010 at
http://www.dataott.org/app/need/show/4
> Almost the entire fleet is GPS equipped.
>
> Each bus in active service has an on-board unit that receives a data packet with its route assignment for that run, called a "block."
>
> The on-board unit constantly compares its location from the GPS output to where it should be at that moment, based on the block data. (Much of this is interpolated because the block has a finite number of location data points.)
>
> Every two minutes, the unit phones home on the Bell network and uploads a data packet that contains the following fields: datetime, block id, route, longitude, latitude, speed, direction, and adherence to the scheduled route, expressed in +/- seconds.
>
> I'm working on a project using these data obtained through FOI but if anyone wants to play with it, let me know. The data set for the past year is about 83 million records.
>
> My understanding from talking to the city people is that they are currently upgrading the system that receives the uploaded data. I don't think there's any chance of an API to the live data until that process is complete.
>
> Glen McGregor Ottawa Citizen

It seems like he already has the aggregate data, and is willing to share it.

Jonathan

[Shawn M. Hooper]

Jonathan,

It sounds like Glen's FOI dataset would have been for a point in time, not ongoing from then.

I've left him a message to discuss. I'll let you know when i hear back from him.

Shawn
Sent from my BlackBerry device on the Rogers Wireless Network

[Shawn M. Hooper]

Hi All,

I spoke with Glen McGregor at the Ottawa Citizen yesterday regarding the
data collection project we'd been discussing. The dataset he has in his
possession covers a one-year period - I believe he said from April 2009
to April 2010. It was received through a FOI request.

He has not received any data newer than that point, so there would be
value in our group creating something for this purpose.

Edward - at this point, it would be interesting to see if that professor
is still interesting in doing something with this data? Do you have
that contact information?

Shawn

[Alex Lougheed]

Howdy all,

I got the professor's contact information. Left his business card at home though, so I can get it to you this evening. He's particularly interested in how natural variables change the system (such as weather, time of day, etc.)

A

--
Alex Lougheed
alougheed.ca  613 700 2358

[Alex Lougheed]

David J. Wright.
dwr...@uottawa.ca
6135625800 (4784)

[Shawn M. Hooper]

Thanks Alex..... guess this project will be on hold until the GPS feed is back.... sigh.

Cheers.
Shawn