{
  "id": 535835,
  "title": "What is the best score in theory?",
  "url": "/competitions/ariel-data-challenge-2024/discussion/535835",
  "author_name": "Timmy Juicehouse",
  "post_date": "2024-09-24T13:59:05.067000",
  "votes": 0,
  "comment_count": 2,
  "views": 0,
  "content": "<p>In my understanding,  the best score is that, predicted values are infinitely close to true values, predicted sigmas approaching zero. </p>\n<p>So better cv is like that almost one-like RMSE and almost zero-like sigmas?</p>\n<p>Am I correct? </p>",
  "messages": [
    {
      "id": 2997393,
      "postDate": "2024-09-24T14:04:54.083Z",
      "content": "<p>1.0 score is achievable with predictions that lie within &lt;1e-5 sigma of true values, because 1e-5 is true-sigma.</p>",
      "rawMarkdown": "1.0 score is achievable with predictions that lie within <1e-5 sigma of true values, because 1e-5 is true-sigma.",
      "votes": 2,
      "replies": [
        {
          "id": 2997417,
          "postDate": "2024-09-24T14:21:49.823Z",
          "content": "<p>Oh, I forgot the true sigma 1e-5.  Thank you, Georgii.</p>",
          "rawMarkdown": "Oh, I forgot the true sigma 1e-5.  Thank you, Georgii."
        }
      ]
    },
    {
      "id": 2997385,
      "postDate": "2024-09-24T13:59:05.067Z",
      "content": "<p>In my understanding,  the best score is that, predicted values are infinitely close to true values, predicted sigmas approaching zero. </p>\n<p>So better cv is like that almost one-like RMSE and almost zero-like sigmas?</p>\n<p>Am I correct? </p>",
      "rawMarkdown": "In my understanding,  the best score is that, predicted values are infinitely close to true values, predicted sigmas approaching zero. \n\nSo better cv is like that almost one-like RMSE and almost zero-like sigmas?\n\nAm I correct? "
    }
  ],
  "comments": [
    {
      "id": 2997393,
      "author_name": "Georgii Aparin",
      "author_url": "",
      "post_date": "2024-09-24T14:04:54.083000",
      "content": "<p>1.0 score is achievable with predictions that lie within &lt;1e-5 sigma of true values, because 1e-5 is true-sigma.</p>",
      "votes": 2,
      "replies": [
        {
          "id": 2997417,
          "author_name": "Timmy Juicehouse",
          "author_url": "",
          "post_date": "2024-09-24T14:21:49.823000",
          "content": "<p>Oh, I forgot the true sigma 1e-5.  Thank you, Georgii.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2997393": "1.0 score is achievable with predictions that lie within <1e-5 sigma of true values, because 1e-5 is true-sigma.",
    "2997385": "In my understanding,  the best score is that, predicted values are infinitely close to true values, predicted sigmas approaching zero. \n\nSo better cv is like that almost one-like RMSE and almost zero-like sigmas?\n\nAm I correct? "
  }
}