{
  "id": 483473,
  "title": "LB score worsens after break",
  "url": "/competitions/home-credit-credit-risk-model-stability/discussion/483473",
  "author_name": "",
  "post_date": "2024-03-12T15:23:26.534170Z",
  "votes": 4,
  "comment_count": 3,
  "views": 0,
  "content": "<p>All my code submissions (with or without retraining) get worse by about 0.01 after the break (test data transformation). It's interesting, is such a relatively high delta specific to my notebooks/realizations or is it more common?</p>",
  "messages": [
    {
      "id": "2693661",
      "postDate": "03/12/2024 15:23:26",
      "content": "<p>All my code submissions (with or without retraining) get worse by about 0.01 after the break (test data transformation). It's interesting, is such a relatively high delta specific to my notebooks/realizations or is it more common?</p>",
      "rawMarkdown": "All my code submissions (with or without retraining) get worse by about 0.01 after the break (test data transformation). It's interesting, is such a relatively high delta specific to my notebooks/realizations or is it more common?",
      "votes": null
    },
    {
      "id": "2693695",
      "postDate": "03/12/2024 15:52:42",
      "content": "<p>The competition hosts have updated the test data. It is a marginal thing. I don't see how the post is relevant to the discussion.</p>",
      "rawMarkdown": "The competition hosts have updated the test data. It is a marginal thing. I don't see how the post is relevant to the discussion.",
      "votes": null
    },
    {
      "id": "2694085",
      "postDate": "03/12/2024 20:25:54",
      "content": "<p>For me it worsened from about 0.01 upto crazy 0.25, this is for a very simple baseline with no metric hacking that didn’t use the removed fields in the test data.</p>",
      "rawMarkdown": "For me it worsened from about 0.01 upto crazy 0.25, this is for a very simple baseline with no metric hacking that didn’t use the removed fields in the test data.",
      "votes": null
    },
    {
      "id": "2695644",
      "postDate": "03/13/2024 19:31:40",
      "content": "<p>Circa 0.05 difference here <a href=\"https://www.kaggle.com/andreynesterov\" target=\"_blank\">@andreynesterov</a> . We are at the top of the leaderboard</p>",
      "rawMarkdown": "Circa 0.05 difference here @andreynesterov . We are at the top of the leaderboard",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2693695,
      "author_name": "matousfamera",
      "author_url": "",
      "post_date": "03/12/2024 15:52:42",
      "content": "<p>The competition hosts have updated the test data. It is a marginal thing. I don't see how the post is relevant to the discussion.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2694085,
      "author_name": "kononenko",
      "author_url": "",
      "post_date": "03/12/2024 20:25:54",
      "content": "<p>For me it worsened from about 0.01 upto crazy 0.25, this is for a very simple baseline with no metric hacking that didn’t use the removed fields in the test data.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2695644,
      "author_name": "octaviograu",
      "author_url": "",
      "post_date": "03/13/2024 19:31:40",
      "content": "<p>Circa 0.05 difference here <a href=\"https://www.kaggle.com/andreynesterov\" target=\"_blank\">@andreynesterov</a> . We are at the top of the leaderboard</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2693661": "All my code submissions (with or without retraining) get worse by about 0.01 after the break (test data transformation). It's interesting, is such a relatively high delta specific to my notebooks/realizations or is it more common?",
    "2693695": "The competition hosts have updated the test data. It is a marginal thing. I don't see how the post is relevant to the discussion.",
    "2694085": "For me it worsened from about 0.01 upto crazy 0.25, this is for a very simple baseline with no metric hacking that didn’t use the removed fields in the test data.",
    "2695644": "Circa 0.05 difference here @andreynesterov . We are at the top of the leaderboard"
  },
  "source": "meta"
}