{
  "id": 20555,
  "title": "Understanding 'is_booking'",
  "url": "/competitions/expedia-hotel-recommendations/discussion/20555",
  "author_name": "",
  "post_date": "2016-04-29T21:54:18.397Z",
  "votes": null,
  "comment_count": 2,
  "views": 462,
  "content": "<p>I'm starting on this comp and wondering about the column - </p>\n\n<pre><code> is_booking     1 if a booking, 0 if a click    tinyint\n</code></pre>\n\n<p>This is present in train, but absent in test. Are we to assume that in test everyone is booking? Or just looking? Is it better to exclude the people who didn't book in train? Or just drop the column from training since that information isn't available in test?</p>",
  "messages": [
    {
      "id": "117627",
      "postDate": "04/29/2016 21:54:18",
      "content": "<p>I'm starting on this comp and wondering about the column - </p>\n\n<pre><code> is_booking     1 if a booking, 0 if a click    tinyint\n</code></pre>\n\n<p>This is present in train, but absent in test. Are we to assume that in test everyone is booking? Or just looking? Is it better to exclude the people who didn't book in train? Or just drop the column from training since that information isn't available in test?</p>",
      "rawMarkdown": "I'm starting on this comp and wondering about the column - \r\n\r\n     is_booking \t1 if a booking, 0 if a click \ttinyint\r\n\r\nThis is present in train, but absent in test. Are we to assume that in test everyone is booking? Or just looking? Is it better to exclude the people who didn't book in train? Or just drop the column from training since that information isn't available in test?",
      "votes": null
    },
    {
      "id": "117632",
      "postDate": "04/29/2016 22:24:34",
      "content": "<p>If you check the competition details, the section about the data it says &quot;Test data only includes booking events&quot;. Therefore, in test, every event is a booking event.</p>\n\n<p>I guess I good way to start is sub setting the train set to only include rows that are booking events, since this is what you want to predict. But maybe the click events has some information that you could use on your model, one would have to explore that to know more.</p>\n\n<p>Good luck.</p>",
      "rawMarkdown": "If you check the competition details, the section about the data it says \"Test data only includes booking events\". Therefore, in test, every event is a booking event.\r\n\r\nI guess I good way to start is sub setting the train set to only include rows that are booking events, since this is what you want to predict. But maybe the click events has some information that you could use on your model, one would have to explore that to know more.\r\n\r\nGood luck.",
      "votes": null
    },
    {
      "id": "117633",
      "postDate": "04/29/2016 22:29:32",
      "content": "<p>Thanks! I don't know how I missed that part.</p>",
      "rawMarkdown": "Thanks! I don't know how I missed that part.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 117632,
      "author_name": "pedromb",
      "author_url": "",
      "post_date": "04/29/2016 22:24:34",
      "content": "<p>If you check the competition details, the section about the data it says &quot;Test data only includes booking events&quot;. Therefore, in test, every event is a booking event.</p>\n\n<p>I guess I good way to start is sub setting the train set to only include rows that are booking events, since this is what you want to predict. But maybe the click events has some information that you could use on your model, one would have to explore that to know more.</p>\n\n<p>Good luck.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 117633,
      "author_name": "piinformatics",
      "author_url": "",
      "post_date": "04/29/2016 22:29:32",
      "content": "<p>Thanks! I don't know how I missed that part.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "117627": "I'm starting on this comp and wondering about the column - \r\n\r\n     is_booking \t1 if a booking, 0 if a click \ttinyint\r\n\r\nThis is present in train, but absent in test. Are we to assume that in test everyone is booking? Or just looking? Is it better to exclude the people who didn't book in train? Or just drop the column from training since that information isn't available in test?",
    "117632": "If you check the competition details, the section about the data it says \"Test data only includes booking events\". Therefore, in test, every event is a booking event.\r\n\r\nI guess I good way to start is sub setting the train set to only include rows that are booking events, since this is what you want to predict. But maybe the click events has some information that you could use on your model, one would have to explore that to know more.\r\n\r\nGood luck.",
    "117633": "Thanks! I don't know how I missed that part."
  },
  "source": "meta"
}