{ "id": 3441, "title": "on submission data format ", "url": "/competitions/flight/discussion/3441", "author_name": "", "post_date": "2012-12-22T09:19:04.997Z", "votes": null, "comment_count": 4, "views": 3427, "content": "

Hi,

\r\n

If the data is supposed to be in "minutes since midnight UTC", how come reference answers sets like "estimated_arrival_benchmark_rev1.csv" have entries that are larger than 24 hours (1440 minutes?).  Perhaps a clearer definition should be in order (e.g.\r\n midnight UTC on the day of scheduled departure). I am adding some ad hoc glue here to my code.

\r\n
flight_history_id,actual_runway_arrival,actual_gate_arrival
280512632,1401,1403
280512634,1407,1406
280512665,1452,1461  <- here
280512670,1441,1446  <- here again
280512721,1395,1409
\r\n
...
\r\n
Cheers,
\r\n
- markku
", "messages": [ { "id": "18411", "postDate": "12/22/2012 09:19:04", "content": "

Hi,

\r\n

If the data is supposed to be in "minutes since midnight UTC", how come reference answers sets like "estimated_arrival_benchmark_rev1.csv" have entries that are larger than 24 hours (1440 minutes?).  Perhaps a clearer definition should be in order (e.g.\r\n midnight UTC on the day of scheduled departure). I am adding some ad hoc glue here to my code.

\r\n
flight_history_id,actual_runway_arrival,actual_gate_arrival
280512632,1401,1403
280512634,1407,1406
280512665,1452,1461  <- here
280512670,1441,1446  <- here again
280512721,1395,1409
\r\n
...
\r\n
Cheers,
\r\n
- markku
", "rawMarkdown": "", "votes": null }, { "id": "18414", "postDate": "12/22/2012 13:39:50", "content": "

I'm pretty sure it's in minutes since midnight of the day the flight departed.

\r\n

So if a flight departs on the 22nd and arrives 06:00:00 on the 23rd the entry would be:

\r\n

(22nd contribution) + (23rd cont)

\r\n

24*60                    + 6*60           = 1800

", "rawMarkdown": "", "votes": null }, { "id": "18416", "postDate": "12/22/2012 13:41:58", "content": "

Yes, minutes since midnight (UTC) of the day that the flight was selected as being in the air.

\r\n

http://www.gequest.com/c/flight/forums/t/3236/requested-clarification-on-predicted-arrival-times

", "rawMarkdown": "", "votes": null }, { "id": "18417", "postDate": "12/22/2012 14:03:22", "content": "

flighthistoryid,departureairportcode,arrivalairportcode,publisheddeparture,publishedarrival,scheduledgatedeparture,scheduledgatearrival,scheduledrunwaydeparture,scheduledrunwayarrival

\r\n

280889846,DEN,DCA,2012-11-20 23:40:00+00:00,2012-11-21 02:52:00+00:00,2012-11-20 23:40:00+00:00,2012-11-21 02:52:00+00:00,2012-11-21 00:10:00+00:00,2012-11-21 03:07:00+00:00

\r\n

the above is a flight that has

\r\n

published_departure = 2012-11-20 23:40:00+00:00

\r\n

scheduledrunwaydeparture 2012-11-21 00:10:00+00:00

\r\n

So we get a new midnight before the flight takes off.

\r\n

here is a flight that has the published_departure missing, so we cannot always use the published departure

\r\n

281215762,ACK,BOS,MISSING,MISSING,2012-11-21 22:30:00+00:00,2012-11-21 23:15:00+00:00,MISSING,MISSING

\r\n

Please clearly specify how the midnight relative to which we make our predictions is calculated.

\r\n

" It'll be the number of minutes past the midnight that started the day that the flight was selected as being in the air. So your 10 minutes should really be 1450 minutes."

\r\n

If for a flight to be selected to be in the air we need an observed runway departure less than the cutoff time that's fine. We just need to nail this down.

", "rawMarkdown": "", "votes": null }, { "id": "18424", "postDate": "12/22/2012 15:53:12", "content": "

The 24 hours we are interested in are from (date)09:00:00+00 to (date+1)09:00:00+00.

\r\n

The midnight reference is (date)00:00:00+00.

\r\n

So, the expected times to report in minutes should be (expected time UTC - midnight reference)*24*60
\r\nwhich can be as high as 1440 + 9*60 = 1980 but should not be less than 540.
\r\nsince expected time UTC can be anywhere in the range of interest (from (date)09:00:00+00 to (date+1)09:00:00+00).

", "rawMarkdown": "", "votes": null } ], "comments": [ { "id": 18414, "author_name": "davidleen", "author_url": "", "post_date": "12/22/2012 13:39:50", "content": "

I'm pretty sure it's in minutes since midnight of the day the flight departed.

\r\n

So if a flight departs on the 22nd and arrives 06:00:00 on the 23rd the entry would be:

\r\n

(22nd contribution) + (23rd cont)

\r\n

24*60                    + 6*60           = 1800

", "votes": null, "replies": [] }, { "id": 18416, "author_name": "bjg9846", "author_url": "", "post_date": "12/22/2012 13:41:58", "content": "

Yes, minutes since midnight (UTC) of the day that the flight was selected as being in the air.

\r\n

http://www.gequest.com/c/flight/forums/t/3236/requested-clarification-on-predicted-arrival-times

", "votes": null, "replies": [] }, { "id": 18417, "author_name": "dxyz46459", "author_url": "", "post_date": "12/22/2012 14:03:22", "content": "

flighthistoryid,departureairportcode,arrivalairportcode,publisheddeparture,publishedarrival,scheduledgatedeparture,scheduledgatearrival,scheduledrunwaydeparture,scheduledrunwayarrival

\r\n

280889846,DEN,DCA,2012-11-20 23:40:00+00:00,2012-11-21 02:52:00+00:00,2012-11-20 23:40:00+00:00,2012-11-21 02:52:00+00:00,2012-11-21 00:10:00+00:00,2012-11-21 03:07:00+00:00

\r\n

the above is a flight that has

\r\n

published_departure = 2012-11-20 23:40:00+00:00

\r\n

scheduledrunwaydeparture 2012-11-21 00:10:00+00:00

\r\n

So we get a new midnight before the flight takes off.

\r\n

here is a flight that has the published_departure missing, so we cannot always use the published departure

\r\n

281215762,ACK,BOS,MISSING,MISSING,2012-11-21 22:30:00+00:00,2012-11-21 23:15:00+00:00,MISSING,MISSING

\r\n

Please clearly specify how the midnight relative to which we make our predictions is calculated.

\r\n

" It'll be the number of minutes past the midnight that started the day that the flight was selected as being in the air. So your 10 minutes should really be 1450 minutes."

\r\n

If for a flight to be selected to be in the air we need an observed runway departure less than the cutoff time that's fine. We just need to nail this down.

", "votes": null, "replies": [] }, { "id": 18424, "author_name": "vadnais", "author_url": "", "post_date": "12/22/2012 15:53:12", "content": "

The 24 hours we are interested in are from (date)09:00:00+00 to (date+1)09:00:00+00.

\r\n

The midnight reference is (date)00:00:00+00.

\r\n

So, the expected times to report in minutes should be (expected time UTC - midnight reference)*24*60
\r\nwhich can be as high as 1440 + 9*60 = 1980 but should not be less than 540.
\r\nsince expected time UTC can be anywhere in the range of interest (from (date)09:00:00+00 to (date+1)09:00:00+00).

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "18411": "", "18414": "", "18416": "", "18417": "", "18424": "" }, "source": "meta" }