{
  "id": 3417,
  "title": "Motivation",
  "url": "/competitions/flight/discussion/3417",
  "author_name": "",
  "post_date": "2012-12-20T15:45:59.973Z",
  "votes": null,
  "comment_count": 2,
  "views": 2933,
  "content": "<blockquote>\r\n<p>The challenge (http://www.gequest.com/c/flight) is to &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 them make flights more efficient\r\n and reliably on time.&quot;</p>\r\n<p>The task is to a) model and update the details of all flights in the 14-day time period up until the cutoff day/time and then to b) predict the runway and gate arrival times of all the remaining&nbsp;flights in the air.</p>\r\n<p>Over a 14-day period there are ~350,000 flights and at the cutoff there are ~26,000 flights to predict for.&nbsp; The volume numbers (350,000 and 26,000) are low to categorize as requiring a scalable solution, and the problem as stated could be solved by for\r\n example, a random forest solution.</p>\r\n<p>A bit baffled between the challenge ('... a usable scalable algorithm that delivers a real-time flight profile ...') and the actual problem which could be solved using a non-scalable and non real-time algorithm.&nbsp; Have I mis-understood the competition challenge\r\n and problem?</p>\r\n</blockquote>",
  "messages": [
    {
      "id": "18273",
      "postDate": "12/20/2012 15:45:59",
      "content": "<blockquote>\r\n<p>The challenge (http://www.gequest.com/c/flight) is to &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 them make flights more efficient\r\n and reliably on time.&quot;</p>\r\n<p>The task is to a) model and update the details of all flights in the 14-day time period up until the cutoff day/time and then to b) predict the runway and gate arrival times of all the remaining&nbsp;flights in the air.</p>\r\n<p>Over a 14-day period there are ~350,000 flights and at the cutoff there are ~26,000 flights to predict for.&nbsp; The volume numbers (350,000 and 26,000) are low to categorize as requiring a scalable solution, and the problem as stated could be solved by for\r\n example, a random forest solution.</p>\r\n<p>A bit baffled between the challenge ('... a usable scalable algorithm that delivers a real-time flight profile ...') and the actual problem which could be solved using a non-scalable and non real-time algorithm.&nbsp; Have I mis-understood the competition challenge\r\n and problem?</p>\r\n</blockquote>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "18623",
      "postDate": "12/26/2012 10:27:02",
      "content": "<p>Hi Ben et al, Any chance of an explanation on this one? Thanks.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "18633",
      "postDate": "12/26/2012 16:10:50",
      "content": "<p>As a rewording of dbv's question, should our model be able to make predictions within, say, 15 minutes on some standard computer?&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 18623,
      "author_name": "intaka",
      "author_url": "",
      "post_date": "12/26/2012 10:27:02",
      "content": "<p>Hi Ben et al, Any chance of an explanation on this one? Thanks.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 18633,
      "author_name": "jay670790",
      "author_url": "",
      "post_date": "12/26/2012 16:10:50",
      "content": "<p>As a rewording of dbv's question, should our model be able to make predictions within, say, 15 minutes on some standard computer?&nbsp;</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "18273": "",
    "18623": "",
    "18633": ""
  },
  "source": "meta"
}