{
  "id": 3541,
  "title": "Real-time system",
  "url": "/competitions/flight/discussion/3541",
  "author_name": "",
  "post_date": "2013-01-06T18:50:45.760Z",
  "votes": null,
  "comment_count": 1,
  "views": 2027,
  "content": "<p>It looks like the benchmark files do not respect the spirit of real time which, I thought, was an important objective of this project.</p>\r\n<p>The flightHistory file represents the status of the flights at the end of the day. Some data from that file (namely actual and scheduled times) are not available throughout the day; they become available in real time as events occur.</p>\r\n<p>Flights are not even known until a RecordCreated event occurs. In the PublicLeaderboardSet, there are two flights (281919861 on 11/28 and 281925858 on 11/29) in the benchmark file which don’t even appear in the flighthistoryevents file before cutoff time.\r\n Since they don’t have published times, it should not<br>\r\nbe possible to make any predictions until an event provides estimated (or scheduled) gate (or runway) departure (or arrival) time. One is enough, but you<br>\r\nneed at least one. Getting scheduled time(s) from the flightHistory file is cheating (and that is the only way to provide a reasonable guess in these cases).</p>\r\n<p>These flight don’t even have a Scheduled (S) status yet, so they cannot meet the rule requiring that they should be Active (A) at cutoff time.</p>\r\n<p>These are extreme cases, but some flights (about 30 a day) got a RecordCreated event, but no scheduled or estimated info before the cutoff time, if they didn’t have Published times in the flightHistory file, they cannot be predicted either and they are most\r\n probably not Active either.</p>\r\n<p>I don’t mind having to predict arrival times for non-Active flights, but the flight should at least have been identified by a RecordCreated event and have Published times or have received Scheduled or Estimated times through a subsequent event.</p>",
  "messages": [
    {
      "id": "19037",
      "postDate": "01/06/2013 18:50:45",
      "content": "<p>It looks like the benchmark files do not respect the spirit of real time which, I thought, was an important objective of this project.</p>\r\n<p>The flightHistory file represents the status of the flights at the end of the day. Some data from that file (namely actual and scheduled times) are not available throughout the day; they become available in real time as events occur.</p>\r\n<p>Flights are not even known until a RecordCreated event occurs. In the PublicLeaderboardSet, there are two flights (281919861 on 11/28 and 281925858 on 11/29) in the benchmark file which don’t even appear in the flighthistoryevents file before cutoff time.\r\n Since they don’t have published times, it should not<br>\r\nbe possible to make any predictions until an event provides estimated (or scheduled) gate (or runway) departure (or arrival) time. One is enough, but you<br>\r\nneed at least one. Getting scheduled time(s) from the flightHistory file is cheating (and that is the only way to provide a reasonable guess in these cases).</p>\r\n<p>These flight don’t even have a Scheduled (S) status yet, so they cannot meet the rule requiring that they should be Active (A) at cutoff time.</p>\r\n<p>These are extreme cases, but some flights (about 30 a day) got a RecordCreated event, but no scheduled or estimated info before the cutoff time, if they didn’t have Published times in the flightHistory file, they cannot be predicted either and they are most\r\n probably not Active either.</p>\r\n<p>I don’t mind having to predict arrival times for non-Active flights, but the flight should at least have been identified by a RecordCreated event and have Published times or have received Scheduled or Estimated times through a subsequent event.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "19056",
      "postDate": "01/07/2013 11:01:50",
      "content": "<p>On a related note, I've previously asked for clarification of the challenge ie.&nbsp;<span>&quot;Use the different data sets found on this page under Get the Data to develop a usable and scalable algorithm that delivers a real-time flight profile to the pilot, helping\r\n them make flights more efficient and reliably on time&quot; (<a href=\"http://www.gequest.com/c/flight\" rel=\"nofollow\" target=\"_blank\">http://www.gequest.com/c/flight</a>)&nbsp;and the actual task which can be solved with a non-scalable and non-real time solution. &nbsp;</span>The\r\n terms 'scalable' and 'real-time&quot; have precise meanings and so far have not seen a clarification from the organisers.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 19056,
      "author_name": "intaka",
      "author_url": "",
      "post_date": "01/07/2013 11:01:50",
      "content": "<p>On a related note, I've previously asked for clarification of the challenge ie.&nbsp;<span>&quot;Use the different data sets found on this page under Get the Data to develop a usable and scalable algorithm that delivers a real-time flight profile to the pilot, helping\r\n them make flights more efficient and reliably on time&quot; (<a href=\"http://www.gequest.com/c/flight\" rel=\"nofollow\" target=\"_blank\">http://www.gequest.com/c/flight</a>)&nbsp;and the actual task which can be solved with a non-scalable and non-real time solution. &nbsp;</span>The\r\n terms 'scalable' and 'real-time&quot; have precise meanings and so far have not seen a clarification from the organisers.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "19037": "",
    "19056": ""
  },
  "source": "meta"
}