{
  "id": 201957,
  "title": "Quick testing new features",
  "url": "/competitions/riiid-test-answer-prediction/discussion/201957",
  "author_name": "",
  "post_date": "2020-12-07T13:53:00.761082300Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Newbie's question about feature engineering. The size of the train set makes exploring value of new features a bit inconvenient. Is there a way to speed up the process? Will it be appropriate to make a conclusion about the new feature's value based on (for example) 10000 samples?</p>",
  "messages": [
    {
      "id": "1105066",
      "postDate": "12/07/2020 13:53:00",
      "content": "<p>Newbie's question about feature engineering. The size of the train set makes exploring value of new features a bit inconvenient. Is there a way to speed up the process? Will it be appropriate to make a conclusion about the new feature's value based on (for example) 10000 samples?</p>",
      "rawMarkdown": "Newbie's question about feature engineering. The size of the train set makes exploring value of new features a bit inconvenient. Is there a way to speed up the process? Will it be appropriate to make a conclusion about the new feature's value based on (for example) 10000 samples?",
      "votes": null
    },
    {
      "id": "1105318",
      "postDate": "12/07/2020 19:09:24",
      "content": "<p>With 100k rows you can make conclusions I think. I tried it.</p>",
      "rawMarkdown": "With 100k rows you can make conclusions I think. I tried it.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1105318,
      "author_name": "bturan19",
      "author_url": "",
      "post_date": "12/07/2020 19:09:24",
      "content": "<p>With 100k rows you can make conclusions I think. I tried it.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1105066": "Newbie's question about feature engineering. The size of the train set makes exploring value of new features a bit inconvenient. Is there a way to speed up the process? Will it be appropriate to make a conclusion about the new feature's value based on (for example) 10000 samples?",
    "1105318": "With 100k rows you can make conclusions I think. I tried it."
  },
  "source": "meta"
}