{
  "id": 54803,
  "title": "Questions about attributed_time, need suggestions please.",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/54803",
  "author_name": "",
  "post_date": "2018-04-17T23:51:52.882642300Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I see that attributed_time is not in the test set, so I am assuming I shouldn't do any feature engineering on it correct? I am confused as to what I should do with this feature, any suggestions?</p>",
  "messages": [
    {
      "id": "315984",
      "postDate": "04/17/2018 23:51:52",
      "content": "<p>I see that attributed_time is not in the test set, so I am assuming I shouldn't do any feature engineering on it correct? I am confused as to what I should do with this feature, any suggestions?</p>",
      "rawMarkdown": "I see that attributed_time is not in the test set, so I am assuming I shouldn't do any feature engineering on it correct? I am confused as to what I should do with this feature, any suggestions?",
      "votes": null
    },
    {
      "id": "315991",
      "postDate": "04/18/2018 00:13:52",
      "content": "<blockquote>\n  <p>I am confused as to what I should do with this feature, any suggestions?</p>\n</blockquote>\n\n<p>That column is a target rather than feature. In fact, a column missing from the test data compared to train is almost always what we are meant to predict.</p>",
      "rawMarkdown": "&gt; I am confused as to what I should do with this feature, any suggestions?\n\nThat column is a target rather than feature. In fact, a column missing from the test data compared to train is almost always what we are meant to predict.",
      "votes": null
    },
    {
      "id": "315992",
      "postDate": "04/18/2018 00:15:31",
      "content": "<blockquote>\n  <p>That column is a target rather than feature. In fact, a column missing from the test data compared to train is almost always what we are meant to predict.</p>\n</blockquote>\n\n<p>Oops, my apology for temporary insanity. I just read the \"attributed\" part rather than \"attributed_time\" that you typed. Most of us are ignoring \"attributed_time\" column.</p>",
      "rawMarkdown": "&gt; That column is a target rather than feature. In fact, a column missing from the test data compared to train is almost always what we are meant to predict.\n\nOops, my apology for temporary insanity. I just read the \"attributed\" part rather than \"attributed_time\" that you typed. Most of us are ignoring \"attributed_time\" column.",
      "votes": null
    },
    {
      "id": "315993",
      "postDate": "04/18/2018 00:18:59",
      "content": "<p>Actually you're right, since only samples have <code>is_attributed == 1</code> will have <code>attributed_time</code> ;)</p>",
      "rawMarkdown": "Actually you're right, since only samples have `is_attributed == 1` will have `attributed_time` ;)",
      "votes": null
    },
    {
      "id": "315994",
      "postDate": "04/18/2018 00:35:38",
      "content": "<p>Thanks guys!</p>",
      "rawMarkdown": "Thanks guys!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 315991,
      "author_name": "tilii7",
      "author_url": "",
      "post_date": "04/18/2018 00:13:52",
      "content": "<blockquote>\n  <p>I am confused as to what I should do with this feature, any suggestions?</p>\n</blockquote>\n\n<p>That column is a target rather than feature. In fact, a column missing from the test data compared to train is almost always what we are meant to predict.</p>",
      "votes": null,
      "replies": [
        {
          "id": 315992,
          "author_name": "tilii7",
          "author_url": "",
          "post_date": "04/18/2018 00:15:31",
          "content": "<blockquote>\n  <p>That column is a target rather than feature. In fact, a column missing from the test data compared to train is almost always what we are meant to predict.</p>\n</blockquote>\n\n<p>Oops, my apology for temporary insanity. I just read the \"attributed\" part rather than \"attributed_time\" that you typed. Most of us are ignoring \"attributed_time\" column.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 315993,
          "author_name": "bangdasun",
          "author_url": "",
          "post_date": "04/18/2018 00:18:59",
          "content": "<p>Actually you're right, since only samples have <code>is_attributed == 1</code> will have <code>attributed_time</code> ;)</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 315994,
          "author_name": "michaelabehsera",
          "author_url": "",
          "post_date": "04/18/2018 00:35:38",
          "content": "<p>Thanks guys!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "315984": "I see that attributed_time is not in the test set, so I am assuming I shouldn't do any feature engineering on it correct? I am confused as to what I should do with this feature, any suggestions?",
    "315991": "&gt; I am confused as to what I should do with this feature, any suggestions?\n\nThat column is a target rather than feature. In fact, a column missing from the test data compared to train is almost always what we are meant to predict.",
    "315992": "&gt; That column is a target rather than feature. In fact, a column missing from the test data compared to train is almost always what we are meant to predict.\n\nOops, my apology for temporary insanity. I just read the \"attributed\" part rather than \"attributed_time\" that you typed. Most of us are ignoring \"attributed_time\" column.",
    "315993": "Actually you're right, since only samples have `is_attributed == 1` will have `attributed_time` ;)",
    "315994": "Thanks guys!"
  },
  "source": "meta"
}