{
  "id": 2794,
  "title": "Looking forward within the test-set?",
  "url": "/competitions/predict-closed-questions-on-stack-overflow/discussion/2794",
  "author_name": "",
  "post_date": "2012-10-02T09:25:31.103Z",
  "votes": null,
  "comment_count": 1,
  "views": 1180,
  "content": "<p>Is it OK to use a model that uses the subsequent (predicted) status of a user's questions as part of its prediction for an earlier question?</p>",
  "messages": [
    {
      "id": "15041",
      "postDate": "10/02/2012 09:25:31",
      "content": "<p>Is it OK to use a model that uses the subsequent (predicted) status of a user's questions as part of its prediction for an earlier question?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "15079",
      "postDate": "10/02/2012 18:32:20",
      "content": "<p>[quote=James;15041]</p>\r\n<p>Is it OK to use a model that uses the subsequent (predicted) status of a user's questions as part of its prediction for an earlier question?</p>\r\n<p>[/quote]No (each question in the test dataset should be predicted independently)</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 15079,
      "author_name": "benhamner",
      "author_url": "",
      "post_date": "10/02/2012 18:32:20",
      "content": "<p>[quote=James;15041]</p>\r\n<p>Is it OK to use a model that uses the subsequent (predicted) status of a user's questions as part of its prediction for an earlier question?</p>\r\n<p>[/quote]No (each question in the test dataset should be predicted independently)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "15041": "",
    "15079": ""
  },
  "source": "meta"
}