2012.05.24 Weekly Check in - atlregional/OpenTripPlanner GitHub Wiki
13:31 <demory> hello everyone
13:31 <novalis_dt> hi
13:31 <novalis_dt> Do you know if Andrew will be in?
13:31 <novalis_dt> Because I broke a test and the CI server didn't mail me.
13:32 <demory> not sure; he was on earlier today wasn't he?
13:32 <novalis_dt> yep
13:32 <demory> in any case he didn't tell me he couldn't make it
13:33 <demory> why don't we go ahead and get started
13:33 <demory> hopefully Andrew will rejoin in a bit
13:34 <demory> my update: have been working on some enhancements for deployer, namely an improved upload interface. also just re-themed the front-end w/ the new OTP graphics: http://deployer.opentripplanner.org/
13:34 <novalis_dt> OK. I've been fixing misc bugs reported by e.g. johngouf and gboudreau.
13:34 <demory> currently working on the presentation for next week's OTP webinar w/ CUTR
13:34 <novalis_dt> I'm going to fix that failing test too
13:38 <demory> not sure if anything is re-usable there
13:38 <mele> Hi James! Grant and I are interns on the project too... for TriMet out in Oregon
13:38 <jcwong86> Hi @mele - nice to meet you guys
13:39 <jcwong86> not sure if we're still doing updates, but I'm playing around with data comparing Capital Bikeshare trips that were taken in Q1 this year with statistics from alternatively planned trips using OTP.
13:39 <FrankP> thank demory...saw that
13:39 <novalis_dt> jcwong86, cool. How are we doing?
13:40 <jcwong86> pretty good. we've got a couple different charts now that are helpful in understanding trips
13:40 <jcwong86> *understanding the kind of trips that bikeshare trips are replacing
13:41 <novalis_dt> Cool.
13:41 <jcwong86> it also helped validate travel times predicted by OTP for bikes. 50% of trips taken actually fall within 2 minutes of the bikeshare trip length
13:41 <jcwong86> 90% within 5 minutes
13:41 <novalis_dt> That is really nice performance, actually.
13:41 <novalis_dt> Are we consistently on one side or the other?
13:42 <novalis_dt> (i.e. too long or too short)
13:42 <jcwong86> *note that this is for fairly short trips, generally 25 min or shorter because of the nature of bikeshare trips
13:43 <jcwong86> its skewed towards predicting too quickly, but we won't conclude that side because people often do things other than rider directly from bike share station to station
13:43 <novalis_dt> Too bad we don't have GPS data for the bikeshare bikes
13:43 <jcwong86> there's terminal time to consider which changes the dynamics. just the fact that those percentages lie within the envelope is pretty good
13:43 <kpw> (we will for nyc!)
13:43 <novalis_dt> kpw, awesome!
13:44 <novalis_dt> Right, we actually do have a terminal time in the code
13:44 <jcwong86> @kpw really? is it just selective or will it transmit realtime?
13:44 <kpw> i think it's available after the trip
13:44 <jcwong86> *the bike times were compared with mode=BICYCLE, not rented bike
13:44 <kpw> don't know details, just keep hearing folks say the nyc bikes have gps
13:44 <novalis_dt> It's assumed to be one minute to pick up and 30 seconds to drop off, for what that's worth.
13:47 <demory> great work on the bike analysis
13:47 <demory> anything else?
13:47 <novalis_dt> Nothing here
13:48 <jcwong86> i'll wrap up the bike stuff and send out a doc with some findings and thoughts. it could probably use some help with any assumptions about OTP algorithms, but i'll pass that around later today or tomorrow.
13:49 <demory> great, thanks James
13:49 <demory> also, next week, that OTP webinar will likely run into our normal check-in time
14:01 <demory> hey abyrd_, we just wrapped up the check-in but if you have anything to report I'll include it in the transcript
14:01 <demory> also I think novalis_dt had a question for you about the CI server
14:01 <abyrd_> hi demory, got stuck in a train.
14:02 <novalis_dt> abyrd_, yeah, I broke the tests recently and didn't get an email
14:02 <abyrd_> I'm just working on realtime stoptime updates and batch queries
14:03 <abyrd_> novalis_dt, yes I saw that too. It is configured to send messages to anyone who breaks the build but is not doing so. Will check if this is a known issue, and if there is an upgrade with a bugfix.