{
  "id": 129249,
  "title": "Pseudo-Labelling",
  "url": "/competitions/cat-in-the-dat-ii/discussion/129249",
  "author_name": "",
  "post_date": "2020-02-06T15:34:53.121831200Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi guys ! Has anyone tried pseudo-labelling ? Precision seems to drop pretty fast on out of fold predictions, so I am wondering if this is worth it ...</p>",
  "messages": [
    {
      "id": "738484",
      "postDate": "02/06/2020 15:34:53",
      "content": "<p>Hi guys ! Has anyone tried pseudo-labelling ? Precision seems to drop pretty fast on out of fold predictions, so I am wondering if this is worth it ...</p>",
      "rawMarkdown": "Hi guys ! Has anyone tried pseudo-labelling ? Precision seems to drop pretty fast on out of fold predictions, so I am wondering if this is worth it ...",
      "votes": null
    },
    {
      "id": "738508",
      "postDate": "02/06/2020 16:16:33",
      "content": "<p>I tried pseudo-labeling with catboost, I could not improve auc .</p>\n\n<p>I added test data with prediction less than 0.01 to training data.</p>",
      "rawMarkdown": "I tried pseudo-labeling with catboost, I could not improve auc .\n\nI added test data with prediction less than 0.01 to training data.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 738508,
      "author_name": "horohoro",
      "author_url": "",
      "post_date": "02/06/2020 16:16:33",
      "content": "<p>I tried pseudo-labeling with catboost, I could not improve auc .</p>\n\n<p>I added test data with prediction less than 0.01 to training data.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "738484": "Hi guys ! Has anyone tried pseudo-labelling ? Precision seems to drop pretty fast on out of fold predictions, so I am wondering if this is worth it ...",
    "738508": "I tried pseudo-labeling with catboost, I could not improve auc .\n\nI added test data with prediction less than 0.01 to training data."
  },
  "source": "meta"
}