{
  "id": 53846,
  "title": "Data for local experiments",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/53846",
  "author_name": "Nikita Varganov",
  "post_date": "2018-04-05T21:08:06.308000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi guys :)\nHow much data do you use to test local hypotheses?</p>\n\n<p>On the one hand, we want to have stable conclusions about the new feature / new idea, but, on the other hand, we want to test many hypotheses and models must be fitted quickly.</p>",
  "messages": [
    {
      "id": 309716,
      "postDate": "2018-04-05T21:08:06.310Z",
      "content": "<p>Hi guys :)\nHow much data do you use to test local hypotheses?</p>\n\n<p>On the one hand, we want to have stable conclusions about the new feature / new idea, but, on the other hand, we want to test many hypotheses and models must be fitted quickly.</p>",
      "rawMarkdown": "Hi guys :)\nHow much data do you use to test local hypotheses?\n\nOn the one hand, we want to have stable conclusions about the new feature / new idea, but, on the other hand, we want to test many hypotheses and models must be fitted quickly."
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "309716": "Hi guys :)\nHow much data do you use to test local hypotheses?\n\nOn the one hand, we want to have stable conclusions about the new feature / new idea, but, on the other hand, we want to test many hypotheses and models must be fitted quickly."
  }
}