{
  "id": 397154,
  "title": "Split training dataset depending on Medication condition",
  "url": "/competitions/tlvmc-parkinsons-freezing-gait-prediction/discussion/397154",
  "author_name": "",
  "post_date": "2023-03-24T11:00:19.877308600Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi, </p>\n<p>I have a doubt about if it would be necessary to split the training dataset in 2 groups. The first one would contain the examples in which a subject has taken the medication (ON condition) and the second one would correspond to those subjects who have not taken the medication (OFF). Otherwise, the feature of Medication should be controlled (taken into account) to avoid a possible confounding situation. Am I wrong? </p>\n<p>Thanks!! </p>",
  "messages": [
    {
      "id": "2195022",
      "postDate": "03/24/2023 11:00:19",
      "content": "<p>Hi, </p>\n<p>I have a doubt about if it would be necessary to split the training dataset in 2 groups. The first one would contain the examples in which a subject has taken the medication (ON condition) and the second one would correspond to those subjects who have not taken the medication (OFF). Otherwise, the feature of Medication should be controlled (taken into account) to avoid a possible confounding situation. Am I wrong? </p>\n<p>Thanks!! </p>",
      "rawMarkdown": "Hi, \n\nI have a doubt about if it would be necessary to split the training dataset in 2 groups. The first one would contain the examples in which a subject has taken the medication (ON condition) and the second one would correspond to those subjects who have not taken the medication (OFF). Otherwise, the feature of Medication should be controlled (taken into account) to avoid a possible confounding situation. Am I wrong? \n\nThanks!!",
      "votes": null
    },
    {
      "id": "2205402",
      "postDate": "04/01/2023 14:00:55",
      "content": "<p>I think the best way to check that will be to run a series of experiments- check if your cv improves by the split.</p>",
      "rawMarkdown": "I think the best way to check that will be to run a series of experiments- check if your cv improves by the split.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2205402,
      "author_name": "avivlevi815",
      "author_url": "",
      "post_date": "04/01/2023 14:00:55",
      "content": "<p>I think the best way to check that will be to run a series of experiments- check if your cv improves by the split.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2195022": "Hi, \n\nI have a doubt about if it would be necessary to split the training dataset in 2 groups. The first one would contain the examples in which a subject has taken the medication (ON condition) and the second one would correspond to those subjects who have not taken the medication (OFF). Otherwise, the feature of Medication should be controlled (taken into account) to avoid a possible confounding situation. Am I wrong? \n\nThanks!!",
    "2205402": "I think the best way to check that will be to run a series of experiments- check if your cv improves by the split."
  },
  "source": "meta"
}