{
  "id": 192752,
  "title": "Guessing 1st place solution AUC",
  "url": "/competitions/riiid-test-answer-prediction/discussion/192752",
  "author_name": "",
  "post_date": "2020-10-23T04:14:17.262703400Z",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Based on the my current understanding of the data and a few models that I have run, the winning solution AUC would be about 0.81xx. </p>\n<p>Any other guesses? :)</p>",
  "messages": [
    {
      "id": "1057837",
      "postDate": "10/23/2020 04:14:17",
      "content": "<p>Based on the my current understanding of the data and a few models that I have run, the winning solution AUC would be about 0.81xx. </p>\n<p>Any other guesses? :)</p>",
      "rawMarkdown": "Based on the my current understanding of the data and a few models that I have run, the winning solution AUC would be about 0.81xx. \n\nAny other guesses? :)",
      "votes": null
    },
    {
      "id": "1057852",
      "postDate": "10/23/2020 04:54:35",
      "content": "<p>Most high scores usually show up in the last 2 weeks of a competition, so any guesses so early (2 months before deadline) will usually be off .. There will be plateaus, then jumps, and incremental (4th digit) improvements.</p>\n<p>Overfitting may give a high score of 0.81xx or more (since Public LB is calculated using only 20% of the data), whereas a robust solution with a \"low\" score on Public LB may end up being the overall winner on the <strong>Private LB</strong> (shake-up!)</p>",
      "rawMarkdown": "Most high scores usually show up in the last 2 weeks of a competition, so any guesses so early (2 months before deadline) will usually be off .. There will be plateaus, then jumps, and incremental (4th digit) improvements.\n\nOverfitting may give a high score of 0.81xx or more (since Public LB is calculated using only 20% of the data), whereas a robust solution with a \"low\" score on Public LB may end up being the overall winner on the **Private LB** (shake-up!)",
      "votes": null
    },
    {
      "id": "1133363",
      "postDate": "12/31/2020 06:29:09",
      "content": "<p>Not that far off for a prediction made in the first week of the competition :)</p>",
      "rawMarkdown": "Not that far off for a prediction made in the first week of the competition :)",
      "votes": null
    },
    {
      "id": "1133491",
      "postDate": "12/31/2020 09:02:04",
      "content": "<p>Prophet 👍👍</p>",
      "rawMarkdown": "Prophet 👍👍",
      "votes": null
    },
    {
      "id": "1133580",
      "postDate": "12/31/2020 10:51:56",
      "content": "<p>Na, just a calculated guess. :)</p>",
      "rawMarkdown": "Na, just a calculated guess. :)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1057852,
      "author_name": "sirishks",
      "author_url": "",
      "post_date": "10/23/2020 04:54:35",
      "content": "<p>Most high scores usually show up in the last 2 weeks of a competition, so any guesses so early (2 months before deadline) will usually be off .. There will be plateaus, then jumps, and incremental (4th digit) improvements.</p>\n<p>Overfitting may give a high score of 0.81xx or more (since Public LB is calculated using only 20% of the data), whereas a robust solution with a \"low\" score on Public LB may end up being the overall winner on the <strong>Private LB</strong> (shake-up!)</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1133363,
      "author_name": "abdurrafae",
      "author_url": "",
      "post_date": "12/31/2020 06:29:09",
      "content": "<p>Not that far off for a prediction made in the first week of the competition :)</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1133491,
      "author_name": "yangxiaoshuai",
      "author_url": "",
      "post_date": "12/31/2020 09:02:04",
      "content": "<p>Prophet 👍👍</p>",
      "votes": null,
      "replies": [
        {
          "id": 1133580,
          "author_name": "abdurrafae",
          "author_url": "",
          "post_date": "12/31/2020 10:51:56",
          "content": "<p>Na, just a calculated guess. :)</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1057837": "Based on the my current understanding of the data and a few models that I have run, the winning solution AUC would be about 0.81xx. \n\nAny other guesses? :)",
    "1057852": "Most high scores usually show up in the last 2 weeks of a competition, so any guesses so early (2 months before deadline) will usually be off .. There will be plateaus, then jumps, and incremental (4th digit) improvements.\n\nOverfitting may give a high score of 0.81xx or more (since Public LB is calculated using only 20% of the data), whereas a robust solution with a \"low\" score on Public LB may end up being the overall winner on the **Private LB** (shake-up!)",
    "1133363": "Not that far off for a prediction made in the first week of the competition :)",
    "1133491": "Prophet 👍👍",
    "1133580": "Na, just a calculated guess. :)"
  },
  "source": "meta"
}