{
  "id": 314460,
  "title": "About the way splitting the dataset",
  "url": "/competitions/happy-whale-and-dolphin/discussion/314460",
  "author_name": "",
  "post_date": "2022-03-22T18:08:37.544772200Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Should we split the dataset by species or individual ids or another rule?</p>",
  "messages": [
    {
      "id": "1731822",
      "postDate": "03/22/2022 18:08:37",
      "content": "<p>Should we split the dataset by species or individual ids or another rule?</p>",
      "rawMarkdown": "Should we split the dataset by species or individual ids or another rule?",
      "votes": null
    },
    {
      "id": "1747231",
      "postDate": "04/06/2022 14:06:42",
      "content": "<p>I'm still working on local CV strategy to prevent overfitting😄.I've tried skfold on individual ids or just random split the trainset.The gap between CV and LB is ~0.08.</p>",
      "rawMarkdown": "I'm still working on local CV strategy to prevent overfitting😄.I've tried skfold on individual ids or just random split the trainset.The gap between CV and LB is ~0.08.",
      "votes": null
    },
    {
      "id": "1747331",
      "postDate": "04/06/2022 15:20:09",
      "content": "<p>I'm still using species split method and also have the slightly similar gap, I wonder if the way splitting method can have much affect the result 🙉</p>",
      "rawMarkdown": "I'm still using species split method and also have the slightly similar gap, I wonder if the way splitting method can have much affect the result 🙉",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1747231,
      "author_name": "jimmysmith1009",
      "author_url": "",
      "post_date": "04/06/2022 14:06:42",
      "content": "<p>I'm still working on local CV strategy to prevent overfitting😄.I've tried skfold on individual ids or just random split the trainset.The gap between CV and LB is ~0.08.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1747331,
          "author_name": "dqhdqmcttdqx",
          "author_url": "",
          "post_date": "04/06/2022 15:20:09",
          "content": "<p>I'm still using species split method and also have the slightly similar gap, I wonder if the way splitting method can have much affect the result 🙉</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1731822": "Should we split the dataset by species or individual ids or another rule?",
    "1747231": "I'm still working on local CV strategy to prevent overfitting😄.I've tried skfold on individual ids or just random split the trainset.The gap between CV and LB is ~0.08.",
    "1747331": "I'm still using species split method and also have the slightly similar gap, I wonder if the way splitting method can have much affect the result 🙉"
  },
  "source": "meta"
}