{
  "id": 552490,
  "title": "key take-aways from this competition",
  "url": "/competitions/child-mind-institute-problematic-internet-use/discussion/552490",
  "author_name": "",
  "post_date": "2024-12-20T00:59:58.208386700Z",
  "votes": 5,
  "comment_count": 7,
  "views": 0,
  "content": "<p>Trust your cv and don't blindly copy the best public notebook! 😂</p>\n<p>My 2 submissions are:</p>\n<ol>\n<li>my best cv notebook. Public score is 0.464 and private score is 0.442.</li>\n<li>a copy of the 0.497 public notebook. Public score is 0.497 and private score is 0.407.</li>\n</ol>\n<p>I believe this competition offers a valuable lesson for those who join near the end and blindly duplicate the best public notebooks (including myself 😅).</p>",
  "messages": [
    {
      "id": "3076442",
      "postDate": "12/20/2024 00:59:58",
      "content": "<p>Trust your cv and don't blindly copy the best public notebook! 😂</p>\n<p>My 2 submissions are:</p>\n<ol>\n<li>my best cv notebook. Public score is 0.464 and private score is 0.442.</li>\n<li>a copy of the 0.497 public notebook. Public score is 0.497 and private score is 0.407.</li>\n</ol>\n<p>I believe this competition offers a valuable lesson for those who join near the end and blindly duplicate the best public notebooks (including myself 😅).</p>",
      "rawMarkdown": "Trust your cv and don't blindly copy the best public notebook! 😂\n\nMy 2 submissions are:\n1. my best cv notebook. Public score is 0.464 and private score is 0.442.\n2. a copy of the 0.497 public notebook. Public score is 0.497 and private score is 0.407.\n\nI believe this competition offers a valuable lesson for those who join near the end and blindly duplicate the best public notebooks (including myself 😅).",
      "votes": null
    },
    {
      "id": "3076597",
      "postDate": "12/20/2024 05:28:31",
      "content": "<p>Hello all,</p>\n<p>My key takeaways are as below-</p>\n<ol>\n<li>Avoid such competitions (best option)</li>\n<li>If you decide to participate, do not spend more than 3-4 days </li>\n<li>Work on a simple model and do not try to tune parameters, thresholds. Use default parameters (they work better than tuned parameters in such cases) </li>\n<li>Make an honest submission with simple models and exit the competition</li>\n<li>Avoid complex approaches in the final submission and rely on very simple submissions and methods</li>\n<li>Rely on luck and don't look back</li>\n</ol>\n<p>Thoughts? Comments?</p>",
      "rawMarkdown": "Hello all,\n\nMy key takeaways are as below-\n1. Avoid such competitions (best option)\n2. If you decide to participate, do not spend more than 3-4 days \n3. Work on a simple model and do not try to tune parameters, thresholds. Use default parameters (they work better than tuned parameters in such cases) \n4. Make an honest submission with simple models and exit the competition\n5. Avoid complex approaches in the final submission and rely on very simple submissions and methods\n6. Rely on luck and don't look back\n\nThoughts? Comments?",
      "votes": null
    },
    {
      "id": "3076684",
      "postDate": "12/20/2024 07:26:34",
      "content": "<p>Totally agree,especially the first two points :)</p>\n<p>I got my best ever rank in this competition, but the total time I spent on this competition was less than 10 minutes. I simply copied two public notebooks and replaced the seed with my favorite numbers — didn’t even look at the data. I feel the silver medal is more of a sarcasm than a glory, as if it was saying: “trust cv my ass, trust random seeds 😂” </p>",
      "rawMarkdown": "Totally agree,especially the first two points :)\n\nI got my best ever rank in this competition, but the total time I spent on this competition was less than 10 minutes. I simply copied two public notebooks and replaced the seed with my favorite numbers — didn’t even look at the data. I feel the silver medal is more of a sarcasm than a glory, as if it was saying: “trust cv my ass, trust random seeds 😂”",
      "votes": null
    },
    {
      "id": "3076686",
      "postDate": "12/20/2024 07:28:18",
      "content": "<p><a href=\"https://www.kaggle.com/shiyili\" target=\"_blank\">@shiyili</a> which lucky kernels did you copy?<br>\nCongratulations for the silver medal!</p>",
      "rawMarkdown": "shiyili which lucky kernels did you copy?\nCongratulations for the silver medal!",
      "votes": null
    },
    {
      "id": "3076742",
      "postDate": "12/20/2024 08:19:47",
      "content": "<p>I am also very curious to see the top score notebooks</p>",
      "rawMarkdown": "I am also very curious to see the top score notebooks",
      "votes": null
    },
    {
      "id": "3076837",
      "postDate": "12/20/2024 10:08:16",
      "content": "<p>Thanks！I luckily chose <code>CMI-PIU: DeepTables NN [CV 0.482]</code> and <code>[3] CMI 🏁 End to End</code> . The authors have done an amazing job, excellent code and documentations. </p>",
      "rawMarkdown": "Thanks！I luckily chose `CMI-PIU: DeepTables NN [CV 0.482]` and `[3] CMI 🏁 End to End` . The authors have done an amazing job, excellent code and documentations.",
      "votes": null
    },
    {
      "id": "3077113",
      "postDate": "12/20/2024 15:17:57",
      "content": "<p>Those takeaways are too accurate lol. This is exactly how I operated. </p>",
      "rawMarkdown": "Those takeaways are too accurate lol. This is exactly how I operated.",
      "votes": null
    },
    {
      "id": "3077218",
      "postDate": "12/20/2024 17:29:22",
      "content": "<p>Thank you for choosing  NN. I actually got a couple of medals too in the past comps for a fork-run-submit. There is a time to pay back to the community.</p>",
      "rawMarkdown": "Thank you for choosing ~~to fly with our airlines~~ NN. I actually got a couple of medals too in the past comps for a fork-run-submit. There is a time to pay back to the community.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3076597,
      "author_name": "ravi20076",
      "author_url": "",
      "post_date": "12/20/2024 05:28:31",
      "content": "<p>Hello all,</p>\n<p>My key takeaways are as below-</p>\n<ol>\n<li>Avoid such competitions (best option)</li>\n<li>If you decide to participate, do not spend more than 3-4 days </li>\n<li>Work on a simple model and do not try to tune parameters, thresholds. Use default parameters (they work better than tuned parameters in such cases) </li>\n<li>Make an honest submission with simple models and exit the competition</li>\n<li>Avoid complex approaches in the final submission and rely on very simple submissions and methods</li>\n<li>Rely on luck and don't look back</li>\n</ol>\n<p>Thoughts? Comments?</p>",
      "votes": null,
      "replies": [
        {
          "id": 3076684,
          "author_name": "shiyili",
          "author_url": "",
          "post_date": "12/20/2024 07:26:34",
          "content": "<p>Totally agree,especially the first two points :)</p>\n<p>I got my best ever rank in this competition, but the total time I spent on this competition was less than 10 minutes. I simply copied two public notebooks and replaced the seed with my favorite numbers — didn’t even look at the data. I feel the silver medal is more of a sarcasm than a glory, as if it was saying: “trust cv my ass, trust random seeds 😂” </p>",
          "votes": null,
          "replies": [
            {
              "id": 3076686,
              "author_name": "ravi20076",
              "author_url": "",
              "post_date": "12/20/2024 07:28:18",
              "content": "<p><a href=\"https://www.kaggle.com/shiyili\" target=\"_blank\">@shiyili</a> which lucky kernels did you copy?<br>\nCongratulations for the silver medal!</p>",
              "votes": null,
              "replies": [
                {
                  "id": 3076837,
                  "author_name": "shiyili",
                  "author_url": "",
                  "post_date": "12/20/2024 10:08:16",
                  "content": "<p>Thanks！I luckily chose <code>CMI-PIU: DeepTables NN [CV 0.482]</code> and <code>[3] CMI 🏁 End to End</code> . The authors have done an amazing job, excellent code and documentations. </p>",
                  "votes": null,
                  "replies": [
                    {
                      "id": 3077218,
                      "author_name": "yekenot",
                      "author_url": "",
                      "post_date": "12/20/2024 17:29:22",
                      "content": "<p>Thank you for choosing  NN. I actually got a couple of medals too in the past comps for a fork-run-submit. There is a time to pay back to the community.</p>",
                      "votes": null,
                      "replies": []
                    }
                  ]
                }
              ]
            },
            {
              "id": 3076742,
              "author_name": "",
              "author_url": "",
              "post_date": "12/20/2024 08:19:47",
              "content": "<p>I am also very curious to see the top score notebooks</p>",
              "votes": null,
              "replies": []
            }
          ]
        },
        {
          "id": 3077113,
          "author_name": "shlomoron",
          "author_url": "",
          "post_date": "12/20/2024 15:17:57",
          "content": "<p>Those takeaways are too accurate lol. This is exactly how I operated. </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3076442": "Trust your cv and don't blindly copy the best public notebook! 😂\n\nMy 2 submissions are:\n1. my best cv notebook. Public score is 0.464 and private score is 0.442.\n2. a copy of the 0.497 public notebook. Public score is 0.497 and private score is 0.407.\n\nI believe this competition offers a valuable lesson for those who join near the end and blindly duplicate the best public notebooks (including myself 😅).",
    "3076597": "Hello all,\n\nMy key takeaways are as below-\n1. Avoid such competitions (best option)\n2. If you decide to participate, do not spend more than 3-4 days \n3. Work on a simple model and do not try to tune parameters, thresholds. Use default parameters (they work better than tuned parameters in such cases) \n4. Make an honest submission with simple models and exit the competition\n5. Avoid complex approaches in the final submission and rely on very simple submissions and methods\n6. Rely on luck and don't look back\n\nThoughts? Comments?",
    "3076684": "Totally agree,especially the first two points :)\n\nI got my best ever rank in this competition, but the total time I spent on this competition was less than 10 minutes. I simply copied two public notebooks and replaced the seed with my favorite numbers — didn’t even look at the data. I feel the silver medal is more of a sarcasm than a glory, as if it was saying: “trust cv my ass, trust random seeds 😂”",
    "3076686": "shiyili which lucky kernels did you copy?\nCongratulations for the silver medal!",
    "3076742": "I am also very curious to see the top score notebooks",
    "3076837": "Thanks！I luckily chose `CMI-PIU: DeepTables NN [CV 0.482]` and `[3] CMI 🏁 End to End` . The authors have done an amazing job, excellent code and documentations.",
    "3077113": "Those takeaways are too accurate lol. This is exactly how I operated.",
    "3077218": "Thank you for choosing ~~to fly with our airlines~~ NN. I actually got a couple of medals too in the past comps for a fork-run-submit. There is a time to pay back to the community."
  },
  "source": "meta"
}