{
  "id": 80488,
  "title": "How Kaggle Team rank 0.70235 at bronze-nothing layer?",
  "url": "/competitions/quora-insincere-questions-classification/discussion/80488",
  "author_name": "",
  "post_date": "2019-02-14T01:47:09.383532700Z",
  "votes": 1,
  "comment_count": 5,
  "views": 0,
  "content": "<p>I have noticed that from 275th to 505th competitors in private leaderboard all get the same score: 0.70235. And many of them only have one submission. That means, these people probabily use the same Forked Kernel and get the same result.\nI just curious about it, how Kaggle Team rank these identity result, By the time of submission? Or just randomly rank them? This leads to some 0.70235 get bronze but other 0.70235 get nothing.</p>",
  "messages": [
    {
      "id": "471066",
      "postDate": "02/14/2019 01:47:09",
      "content": "<p>I have noticed that from 275th to 505th competitors in private leaderboard all get the same score: 0.70235. And many of them only have one submission. That means, these people probabily use the same Forked Kernel and get the same result.\nI just curious about it, how Kaggle Team rank these identity result, By the time of submission? Or just randomly rank them? This leads to some 0.70235 get bronze but other 0.70235 get nothing.</p>",
      "rawMarkdown": "I have noticed that from 275th to 505th competitors in private leaderboard all get the same score: 0.70235. And many of them only have one submission. That means, these people probabily use the same Forked Kernel and get the same result.\nI just curious about it, how Kaggle Team rank these identity result, By the time of submission? Or just randomly rank them? This leads to some 0.70235 get bronze but other 0.70235 get nothing.",
      "votes": null
    },
    {
      "id": "471069",
      "postDate": "02/14/2019 01:55:45",
      "content": "<p>Only four significant digits are shown on the leaderboard, but in fact, as long as the number of significant digits is sufficient, it can always be separated.</p>",
      "rawMarkdown": "Only four significant digits are shown on the leaderboard, but in fact, as long as the number of significant digits is sufficient, it can always be separated.",
      "votes": null
    },
    {
      "id": "471077",
      "postDate": "02/14/2019 02:06:45",
      "content": "<p>However, you cannot find a second scene in private LB that over 200 competitors get an identical 5-digit result. And some 0.70235s indicated the name of Forked Kernels, as you can see in private LB. Anyway, only Kaggle Team knows the precise results of those 0.70235s.</p>",
      "rawMarkdown": "However, you cannot find a second scene in private LB that over 200 competitors get an identical 5-digit result. And some 0.70235s indicated the name of Forked Kernels, as you can see in private LB. Anyway, only Kaggle Team knows the precise results of those 0.70235s.",
      "votes": null
    },
    {
      "id": "471078",
      "postDate": "02/14/2019 02:09:56",
      "content": "<p><a href=\"/dilapsky\">@dilapsky</a>\nwe need to trust Kaggle Term.</p>",
      "rawMarkdown": "dilapsky\nwe need to trust Kaggle Term.",
      "votes": null
    },
    {
      "id": "471176",
      "postDate": "02/14/2019 05:50:34",
      "content": "<p>I also think that these are forked public kernel. I am wondering that best public kernel has a cv score of 0.68 and 0.7 in public LB. Now they are scoring 0.70235 in private part which makes me to think about line \" always trust your CV \"</p>",
      "rawMarkdown": "I also think that these are forked public kernel. I am wondering that best public kernel has a cv score of 0.68 and 0.7 in public LB. Now they are scoring 0.70235 in private part which makes me to think about line \" always trust your CV \"",
      "votes": null
    },
    {
      "id": "471184",
      "postDate": "02/14/2019 06:02:42",
      "content": "<p>Great Question! The ranking order is by time of submission during Stage 1.</p>",
      "rawMarkdown": "Great Question! The ranking order is by time of submission during Stage 1.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 471069,
      "author_name": "xiaobai1123q",
      "author_url": "",
      "post_date": "02/14/2019 01:55:45",
      "content": "<p>Only four significant digits are shown on the leaderboard, but in fact, as long as the number of significant digits is sufficient, it can always be separated.</p>",
      "votes": null,
      "replies": [
        {
          "id": 471077,
          "author_name": "dilapsky",
          "author_url": "",
          "post_date": "02/14/2019 02:06:45",
          "content": "<p>However, you cannot find a second scene in private LB that over 200 competitors get an identical 5-digit result. And some 0.70235s indicated the name of Forked Kernels, as you can see in private LB. Anyway, only Kaggle Team knows the precise results of those 0.70235s.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 471078,
          "author_name": "xiaobai1123q",
          "author_url": "",
          "post_date": "02/14/2019 02:09:56",
          "content": "<p><a href=\"/dilapsky\">@dilapsky</a>\nwe need to trust Kaggle Term.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 471176,
      "author_name": "suchith0312",
      "author_url": "",
      "post_date": "02/14/2019 05:50:34",
      "content": "<p>I also think that these are forked public kernel. I am wondering that best public kernel has a cv score of 0.68 and 0.7 in public LB. Now they are scoring 0.70235 in private part which makes me to think about line \" always trust your CV \"</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 471184,
      "author_name": "inversion",
      "author_url": "",
      "post_date": "02/14/2019 06:02:42",
      "content": "<p>Great Question! The ranking order is by time of submission during Stage 1.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "471066": "I have noticed that from 275th to 505th competitors in private leaderboard all get the same score: 0.70235. And many of them only have one submission. That means, these people probabily use the same Forked Kernel and get the same result.\nI just curious about it, how Kaggle Team rank these identity result, By the time of submission? Or just randomly rank them? This leads to some 0.70235 get bronze but other 0.70235 get nothing.",
    "471069": "Only four significant digits are shown on the leaderboard, but in fact, as long as the number of significant digits is sufficient, it can always be separated.",
    "471077": "However, you cannot find a second scene in private LB that over 200 competitors get an identical 5-digit result. And some 0.70235s indicated the name of Forked Kernels, as you can see in private LB. Anyway, only Kaggle Team knows the precise results of those 0.70235s.",
    "471078": "dilapsky\nwe need to trust Kaggle Term.",
    "471176": "I also think that these are forked public kernel. I am wondering that best public kernel has a cv score of 0.68 and 0.7 in public LB. Now they are scoring 0.70235 in private part which makes me to think about line \" always trust your CV \"",
    "471184": "Great Question! The ranking order is by time of submission during Stage 1."
  },
  "source": "meta"
}