{
  "id": 208297,
  "title": "Public LB reshuffled?",
  "url": "/competitions/cassava-leaf-disease-classification/discussion/208297",
  "author_name": "Junyi Ng",
  "post_date": "2021-01-02T18:02:19.556000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I read the statement regarding the leak and although it was mentioned that no changes were made to the test data, were the public/private test sets reshuffled (or any other changes)?</p>\n<p>Reason being I've been getting higher CV but lower public LB scores and this phenomena only happened since the leak was announced. It's unlikely for me to have data leakage as I only implemented heavier regularization to get higher CV. Is this just me?</p>",
  "messages": [
    {
      "id": 1136087,
      "postDate": "2021-01-02T18:02:19.557Z",
      "content": "<p>I read the statement regarding the leak and although it was mentioned that no changes were made to the test data, were the public/private test sets reshuffled (or any other changes)?</p>\n<p>Reason being I've been getting higher CV but lower public LB scores and this phenomena only happened since the leak was announced. It's unlikely for me to have data leakage as I only implemented heavier regularization to get higher CV. Is this just me?</p>",
      "rawMarkdown": "I read the statement regarding the leak and although it was mentioned that no changes were made to the test data, were the public/private test sets reshuffled (or any other changes)?\n\nReason being I've been getting higher CV but lower public LB scores and this phenomena only happened since the leak was announced. It's unlikely for me to have data leakage as I only implemented heavier regularization to get higher CV. Is this just me?"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1136087": "I read the statement regarding the leak and although it was mentioned that no changes were made to the test data, were the public/private test sets reshuffled (or any other changes)?\n\nReason being I've been getting higher CV but lower public LB scores and this phenomena only happened since the leak was announced. It's unlikely for me to have data leakage as I only implemented heavier regularization to get higher CV. Is this just me?"
  }
}