{
  "id": 39661,
  "title": "if we ensemble the results of all kagglers with LB>0.9969, what will happen?",
  "url": "/competitions/carvana-image-masking-challenge/discussion/39661",
  "author_name": "",
  "post_date": "2017-09-18T18:45:34.626651200Z",
  "votes": 11,
  "comment_count": 9,
  "views": 0,
  "content": "<p>With the results so close, i wonder what will happen if we ensemble all the results of LB&gt;0.9969 together? I am considering making a very large team for this experiment. <strong>If you are interested, please drop your name here.</strong> </p>\n\n<p>Let's see how many kagglers are interested berfore we make the next move.</p>\n\n<p>NOTE:</p>\n\n<ul>\n<li><p>we will form a team if there are more than 10 interested participiants.</p></li>\n<li><p>we need each model to have similiar scores for ensemble to work. So we need them to be LB&gt;=0.9969  </p></li>\n<li><p>if we do win a prize, it will be split equally</p></li>\n<li><p>each team can only submit 5 enteries per day. With such a big team, not all members can submit. we need to work that out later.</p></li>\n<li><p>i checkedthe rule, \"There is no maximum team size.\"\n<a href=\"https://www.kaggle.com/c/carvana-image-masking-challenge/rules\">https://www.kaggle.com/c/carvana-image-masking-challenge/rules</a></p></li>\n</ul>\n\n<p>Good Luck to all!</p>",
  "messages": [
    {
      "id": "222413",
      "postDate": "09/18/2017 18:45:34",
      "content": "<p>With the results so close, i wonder what will happen if we ensemble all the results of LB&gt;0.9969 together? I am considering making a very large team for this experiment. <strong>If you are interested, please drop your name here.</strong> </p>\n\n<p>Let's see how many kagglers are interested berfore we make the next move.</p>\n\n<p>NOTE:</p>\n\n<ul>\n<li><p>we will form a team if there are more than 10 interested participiants.</p></li>\n<li><p>we need each model to have similiar scores for ensemble to work. So we need them to be LB&gt;=0.9969  </p></li>\n<li><p>if we do win a prize, it will be split equally</p></li>\n<li><p>each team can only submit 5 enteries per day. With such a big team, not all members can submit. we need to work that out later.</p></li>\n<li><p>i checkedthe rule, \"There is no maximum team size.\"\n<a href=\"https://www.kaggle.com/c/carvana-image-masking-challenge/rules\">https://www.kaggle.com/c/carvana-image-masking-challenge/rules</a></p></li>\n</ul>\n\n<p>Good Luck to all!</p>",
      "rawMarkdown": "With the results so close, i wonder what will happen if we ensemble all the results of LB&gt;0.9969 together? I am considering making a very large team for this experiment. **If you are interested, please drop your name here.** \n\nLet's see how many kagglers are interested berfore we make the next move.\n\nNOTE:\n\n- we will form a team if there are more than 10 interested participiants.\n\n- we need each model to have similiar scores for ensemble to work. So we need them to be LB&gt;=0.9969  \n\n- if we do win a prize, it will be split equally\n\n- each team can only submit 5 enteries per day. With such a big team, not all members can submit. we need to work that out later.\n\n- i checkedthe rule, \"There is no maximum team size.\"\nhttps://www.kaggle.com/c/carvana-image-masking-challenge/rules\n\nGood Luck to all!",
      "votes": null
    },
    {
      "id": "222416",
      "postDate": "09/18/2017 19:09:31",
      "content": "<p>Well played. Have you read \"Ready Player One\"? Name your team \"Sixers\"</p>",
      "rawMarkdown": "Well played. Have you read \"Ready Player One\"? Name your team \"Sixers\"",
      "votes": null
    },
    {
      "id": "222417",
      "postDate": "09/18/2017 19:14:02",
      "content": "<p>It will probably improve my rank by 9 so go ahead :D\nJust be aware that you have two days before team merger deadline.</p>",
      "rawMarkdown": "It will probably improve my rank by 9 so go ahead :D\nJust be aware that you have two days before team merger deadline.",
      "votes": null
    },
    {
      "id": "222418",
      "postDate": "09/18/2017 19:14:43",
      "content": "<p>Wait for ur 0.9969 Heng :D</p>\n\n<p>Seems pretty good idea</p>",
      "rawMarkdown": "Wait for ur 0.9969 Heng :D\n\nSeems pretty good idea",
      "votes": null
    },
    {
      "id": "222458",
      "postDate": "09/18/2017 22:16:28",
      "content": "<p>you can play with different ensemble weighting methods to potentially enhance your score as well.</p>",
      "rawMarkdown": "you can play with different ensemble weighting methods to potentially enhance your score as well.",
      "votes": null
    },
    {
      "id": "222480",
      "postDate": "09/19/2017 01:06:36",
      "content": "<p>Sounds good. You can count us in. Good luck. </p>",
      "rawMarkdown": "Sounds good. You can count us in. Good luck.",
      "votes": null
    },
    {
      "id": "222564",
      "postDate": "09/19/2017 07:46:27",
      "content": "<p>you can count me in</p>",
      "rawMarkdown": "you can count me in",
      "votes": null
    },
    {
      "id": "222595",
      "postDate": "09/19/2017 10:48:49",
      "content": "<p>As a beginner in deep learning ,what i can do is to give you best wish</p>",
      "rawMarkdown": "As a beginner in deep learning ,what i can do is to give you best wish",
      "votes": null
    },
    {
      "id": "222656",
      "postDate": "09/19/2017 16:00:38",
      "content": "<p>Seems that this idea hasn't attracted much interest, but i will wait till end of the merger deadline.</p>\n\n<p>But i am still interested in the idea. If I cannot do it for the competition, i would like to do it after the competition. So I will ask people to \"donate\" their csv file after the competition and see what results we really can get if we really ensemble \"ALL\" the 0.9969, 0.9968, etc results.</p>",
      "rawMarkdown": "Seems that this idea hasn't attracted much interest, but i will wait till end of the merger deadline.\n\nBut i am still interested in the idea. If I cannot do it for the competition, i would like to do it after the competition. So I will ask people to \"donate\" their csv file after the competition and see what results we really can get if we really ensemble \"ALL\" the 0.9969, 0.9968, etc results.",
      "votes": null
    },
    {
      "id": "222749",
      "postDate": "09/19/2017 21:34:37",
      "content": "<p>It looks like x5 is way ahead of you ;-)</p>",
      "rawMarkdown": "It looks like x5 is way ahead of you ;-)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 222416,
      "author_name": "antorsae",
      "author_url": "",
      "post_date": "09/18/2017 19:09:31",
      "content": "<p>Well played. Have you read \"Ready Player One\"? Name your team \"Sixers\"</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 222417,
      "author_name": "gaborfodor",
      "author_url": "",
      "post_date": "09/18/2017 19:14:02",
      "content": "<p>It will probably improve my rank by 9 so go ahead :D\nJust be aware that you have two days before team merger deadline.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 222418,
      "author_name": "raalsky",
      "author_url": "",
      "post_date": "09/18/2017 19:14:43",
      "content": "<p>Wait for ur 0.9969 Heng :D</p>\n\n<p>Seems pretty good idea</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 222458,
      "author_name": "dongxu027",
      "author_url": "",
      "post_date": "09/18/2017 22:16:28",
      "content": "<p>you can play with different ensemble weighting methods to potentially enhance your score as well.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 222480,
      "author_name": "yangxulei",
      "author_url": "",
      "post_date": "09/19/2017 01:06:36",
      "content": "<p>Sounds good. You can count us in. Good luck. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 222564,
      "author_name": "anuragreddygv323",
      "author_url": "",
      "post_date": "09/19/2017 07:46:27",
      "content": "<p>you can count me in</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 222595,
      "author_name": "cnbest",
      "author_url": "",
      "post_date": "09/19/2017 10:48:49",
      "content": "<p>As a beginner in deep learning ,what i can do is to give you best wish</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 222656,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "09/19/2017 16:00:38",
      "content": "<p>Seems that this idea hasn't attracted much interest, but i will wait till end of the merger deadline.</p>\n\n<p>But i am still interested in the idea. If I cannot do it for the competition, i would like to do it after the competition. So I will ask people to \"donate\" their csv file after the competition and see what results we really can get if we really ensemble \"ALL\" the 0.9969, 0.9968, etc results.</p>",
      "votes": null,
      "replies": [
        {
          "id": 222749,
          "author_name": "stevenknguyen",
          "author_url": "",
          "post_date": "09/19/2017 21:34:37",
          "content": "<p>It looks like x5 is way ahead of you ;-)</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "222413": "With the results so close, i wonder what will happen if we ensemble all the results of LB&gt;0.9969 together? I am considering making a very large team for this experiment. **If you are interested, please drop your name here.** \n\nLet's see how many kagglers are interested berfore we make the next move.\n\nNOTE:\n\n- we will form a team if there are more than 10 interested participiants.\n\n- we need each model to have similiar scores for ensemble to work. So we need them to be LB&gt;=0.9969  \n\n- if we do win a prize, it will be split equally\n\n- each team can only submit 5 enteries per day. With such a big team, not all members can submit. we need to work that out later.\n\n- i checkedthe rule, \"There is no maximum team size.\"\nhttps://www.kaggle.com/c/carvana-image-masking-challenge/rules\n\nGood Luck to all!",
    "222416": "Well played. Have you read \"Ready Player One\"? Name your team \"Sixers\"",
    "222417": "It will probably improve my rank by 9 so go ahead :D\nJust be aware that you have two days before team merger deadline.",
    "222418": "Wait for ur 0.9969 Heng :D\n\nSeems pretty good idea",
    "222458": "you can play with different ensemble weighting methods to potentially enhance your score as well.",
    "222480": "Sounds good. You can count us in. Good luck.",
    "222564": "you can count me in",
    "222595": "As a beginner in deep learning ,what i can do is to give you best wish",
    "222656": "Seems that this idea hasn't attracted much interest, but i will wait till end of the merger deadline.\n\nBut i am still interested in the idea. If I cannot do it for the competition, i would like to do it after the competition. So I will ask people to \"donate\" their csv file after the competition and see what results we really can get if we really ensemble \"ALL\" the 0.9969, 0.9968, etc results.",
    "222749": "It looks like x5 is way ahead of you ;-)"
  },
  "source": "meta"
}