{
  "id": 121823,
  "title": "Meaningful Final LB",
  "url": "/competitions/ashrae-energy-prediction/discussion/121823",
  "author_name": "",
  "post_date": "2019-12-15T19:55:17.517958300Z",
  "votes": 3,
  "comment_count": 4,
  "views": 0,
  "content": "<blockquote>\n  <p>To be clear, we have reason to believe that enough of the sites are unavailable online that there will be a meaningful final leaderboard. <a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/117357\">Source</a></p>\n</blockquote>\n\n<p>If sites 0,1,2,4,15 are removed from final private LB, there should only be roughly 6 million data points or 65.4% of the 22% original private LB. Said differently, private LB would be 14.4% of original full test size.</p>",
  "messages": [
    {
      "id": "695899",
      "postDate": "12/15/2019 19:55:17",
      "content": "<blockquote>\n  <p>To be clear, we have reason to believe that enough of the sites are unavailable online that there will be a meaningful final leaderboard. <a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/117357\">Source</a></p>\n</blockquote>\n\n<p>If sites 0,1,2,4,15 are removed from final private LB, there should only be roughly 6 million data points or 65.4% of the 22% original private LB. Said differently, private LB would be 14.4% of original full test size.</p>",
      "rawMarkdown": "&gt; To be clear, we have reason to believe that enough of the sites are unavailable online that there will be a meaningful final leaderboard. [Source](https://www.kaggle.com/c/ashrae-energy-prediction/discussion/117357)\n\nIf sites 0,1,2,4,15 are removed from final private LB, there should only be roughly 6 million data points or 65.4% of the 22% original private LB. Said differently, private LB would be 14.4% of original full test size.",
      "votes": null
    },
    {
      "id": "695933",
      "postDate": "12/15/2019 21:56:00",
      "content": "<p>sites 0,1,2,4,15 are the datasets that were disclosed in different public notebooks. I m guessing there are more sites that were downloaded and will be removed.</p>",
      "rawMarkdown": "sites 0,1,2,4,15 are the datasets that were disclosed in different public notebooks. I m guessing there are more sites that were downloaded and will be removed.",
      "votes": null
    },
    {
      "id": "695959",
      "postDate": "12/15/2019 23:32:56",
      "content": "<p>What do you think about site_id=3 ?</p>",
      "rawMarkdown": "What do you think about site_id=3 ?",
      "votes": null
    },
    {
      "id": "696116",
      "postDate": "12/16/2019 06:23:31",
      "content": "<blockquote>\n  <p>roughly 6 million data points</p>\n</blockquote>\n\n<p>This is more than enough.</p>",
      "rawMarkdown": "&gt; roughly 6 million data points\n\nThis is more than enough.",
      "votes": null
    },
    {
      "id": "699599",
      "postDate": "12/20/2019 16:16:20",
      "content": "<p><a href=\"/kyosato\">@kyosato</a> seems like it was compromised as well. After removing sites 0,1,2,3,4,15, the private LB comes down to 11.86%</p>",
      "rawMarkdown": "kyosato seems like it was compromised as well. After removing sites 0,1,2,3,4,15, the private LB comes down to 11.86%",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 695933,
      "author_name": "chabir",
      "author_url": "",
      "post_date": "12/15/2019 21:56:00",
      "content": "<p>sites 0,1,2,4,15 are the datasets that were disclosed in different public notebooks. I m guessing there are more sites that were downloaded and will be removed.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 695959,
      "author_name": "kyosato",
      "author_url": "",
      "post_date": "12/15/2019 23:32:56",
      "content": "<p>What do you think about site_id=3 ?</p>",
      "votes": null,
      "replies": [
        {
          "id": 699599,
          "author_name": "teeyee314",
          "author_url": "",
          "post_date": "12/20/2019 16:16:20",
          "content": "<p><a href=\"/kyosato\">@kyosato</a> seems like it was compromised as well. After removing sites 0,1,2,3,4,15, the private LB comes down to 11.86%</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 696116,
      "author_name": "nroman",
      "author_url": "",
      "post_date": "12/16/2019 06:23:31",
      "content": "<blockquote>\n  <p>roughly 6 million data points</p>\n</blockquote>\n\n<p>This is more than enough.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "695899": "&gt; To be clear, we have reason to believe that enough of the sites are unavailable online that there will be a meaningful final leaderboard. [Source](https://www.kaggle.com/c/ashrae-energy-prediction/discussion/117357)\n\nIf sites 0,1,2,4,15 are removed from final private LB, there should only be roughly 6 million data points or 65.4% of the 22% original private LB. Said differently, private LB would be 14.4% of original full test size.",
    "695933": "sites 0,1,2,4,15 are the datasets that were disclosed in different public notebooks. I m guessing there are more sites that were downloaded and will be removed.",
    "695959": "What do you think about site_id=3 ?",
    "696116": "&gt; roughly 6 million data points\n\nThis is more than enough.",
    "699599": "kyosato seems like it was compromised as well. After removing sites 0,1,2,3,4,15, the private LB comes down to 11.86%"
  },
  "source": "meta"
}