{
  "id": 21012,
  "title": "I think I need an overfitter now",
  "url": "/competitions/avito-duplicate-ads-detection/discussion/21012",
  "author_name": "",
  "post_date": "2016-05-17T11:39:35.743Z",
  "votes": null,
  "comment_count": 2,
  "views": 566,
  "content": "<p>Hi,</p>\n\n<p>Right now, with <strong>36 features</strong> and without much overfitting I have achieved eval roc 0.925023 and <strong>LB roc 0.87993</strong>. Currently 7th but falling behind every hour.</p>\n\n<p>I think this time I need <strong>an official overfitter</strong> ;) who can help me go above my current benchmark. </p>\n\n<p>Interested folks can PM me. </p>\n\n<p>P.S. It's better if you have scored good score with your own features apart from publicly provided scripts. </p>",
  "messages": [
    {
      "id": "120314",
      "postDate": "05/17/2016 11:39:35",
      "content": "<p>Hi,</p>\n\n<p>Right now, with <strong>36 features</strong> and without much overfitting I have achieved eval roc 0.925023 and <strong>LB roc 0.87993</strong>. Currently 7th but falling behind every hour.</p>\n\n<p>I think this time I need <strong>an official overfitter</strong> ;) who can help me go above my current benchmark. </p>\n\n<p>Interested folks can PM me. </p>\n\n<p>P.S. It's better if you have scored good score with your own features apart from publicly provided scripts. </p>",
      "rawMarkdown": "Hi,\r\n\r\nRight now, with **36 features** and without much overfitting I have achieved eval roc 0.925023 and **LB roc 0.87993**. Currently 7th but falling behind every hour.\r\n\r\nI think this time I need **an official overfitter** ;) who can help me go above my current benchmark. \r\n\r\nInterested folks can PM me. \r\n\r\nP.S. It's better if you have scored good score with your own features apart from publicly provided scripts.",
      "votes": null
    },
    {
      "id": "120319",
      "postDate": "05/17/2016 12:05:38",
      "content": "",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "120374",
      "postDate": "05/17/2016 20:53:14",
      "content": "<p>I heard this one is pretty good at it:\n<a href=\"https://www.kaggle.com/mikeskim\">https://www.kaggle.com/mikeskim</a></p>",
      "rawMarkdown": "I heard this one is pretty good at it:\r\nhttps://www.kaggle.com/mikeskim",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 120319,
      "author_name": "kelexu",
      "author_url": "",
      "post_date": "05/17/2016 12:05:38",
      "content": "",
      "votes": null,
      "replies": []
    },
    {
      "id": 120374,
      "author_name": "fernandoprocy",
      "author_url": "",
      "post_date": "05/17/2016 20:53:14",
      "content": "<p>I heard this one is pretty good at it:\n<a href=\"https://www.kaggle.com/mikeskim\">https://www.kaggle.com/mikeskim</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "120314": "Hi,\r\n\r\nRight now, with **36 features** and without much overfitting I have achieved eval roc 0.925023 and **LB roc 0.87993**. Currently 7th but falling behind every hour.\r\n\r\nI think this time I need **an official overfitter** ;) who can help me go above my current benchmark. \r\n\r\nInterested folks can PM me. \r\n\r\nP.S. It's better if you have scored good score with your own features apart from publicly provided scripts.",
    "120319": "",
    "120374": "I heard this one is pretty good at it:\r\nhttps://www.kaggle.com/mikeskim"
  },
  "source": "meta"
}