{
  "id": 5956,
  "title": "How is the unknown test data cleaned for submission testing?",
  "url": "/competitions/expedia-personalized-sort/discussion/5956",
  "author_name": "",
  "post_date": "2013-10-04T01:41:57.673Z",
  "votes": null,
  "comment_count": 3,
  "views": 1146,
  "content": "<p>Erroneous question deleted - sorry about that. Thanks again to TotalNovice for pointing out my misunderstanding.</p>",
  "messages": [
    {
      "id": "32007",
      "postDate": "10/04/2013 01:41:57",
      "content": "<p>Erroneous question deleted - sorry about that. Thanks again to TotalNovice for pointing out my misunderstanding.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "32008",
      "postDate": "10/04/2013 01:57:00",
      "content": "",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "32011",
      "postDate": "10/04/2013 02:38:01",
      "content": "<p>Not sure I understand exactly, but your submission is going to be a bunch of srch_ids and prop_ids. You have the full set of results to rank that will be used for final scoring. The admins have all of the data for which hotels were clicked and which were booked, so they can determine the ideal ordering. In a sense, they don't care what the values of any of the columns other than srch_id, prop_id, booking_bool, and a click_bool, and for none of these will there ever be an NA or NaN in their data. All you submit are the srch_ids and prop_ids, and it's up to you to decide what to do with the other data. Your score is computed irrespective of any NAs. Sorry if I misunderstand what you're asking. Maybe where you're getting confused is with regard to the 25% and 75%. The test.csv contains <span style=\"text-decoration: underline\">all</span> the searches you need to rank. The leaderboard only shows your score on 25% of those searches, and your final score is computed on the 100%</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "32013",
      "postDate": "10/04/2013 03:22:59",
      "content": "<p>TotalNovice: You are right. Thanks! I Guess I had a brain fart there&nbsp; :-)</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 32008,
      "author_name": "robreynoldson",
      "author_url": "",
      "post_date": "10/04/2013 01:57:00",
      "content": "",
      "votes": null,
      "replies": []
    },
    {
      "id": 32011,
      "author_name": "totalnovice",
      "author_url": "",
      "post_date": "10/04/2013 02:38:01",
      "content": "<p>Not sure I understand exactly, but your submission is going to be a bunch of srch_ids and prop_ids. You have the full set of results to rank that will be used for final scoring. The admins have all of the data for which hotels were clicked and which were booked, so they can determine the ideal ordering. In a sense, they don't care what the values of any of the columns other than srch_id, prop_id, booking_bool, and a click_bool, and for none of these will there ever be an NA or NaN in their data. All you submit are the srch_ids and prop_ids, and it's up to you to decide what to do with the other data. Your score is computed irrespective of any NAs. Sorry if I misunderstand what you're asking. Maybe where you're getting confused is with regard to the 25% and 75%. The test.csv contains <span style=\"text-decoration: underline\">all</span> the searches you need to rank. The leaderboard only shows your score on 25% of those searches, and your final score is computed on the 100%</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 32013,
      "author_name": "robreynoldson",
      "author_url": "",
      "post_date": "10/04/2013 03:22:59",
      "content": "<p>TotalNovice: You are right. Thanks! I Guess I had a brain fart there&nbsp; :-)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "32007": "",
    "32008": "",
    "32011": "",
    "32013": ""
  },
  "source": "meta"
}