{
  "id": 54284,
  "title": "Do not specialize on n:n rows",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/54284",
  "author_name": "mezoganet",
  "post_date": "2018-04-11T18:13:37.023000",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Be aware about those n:n rows that could lead to overfitting... Be somewhat generalizing, and be confident on your LB... That's my advice.</p>\n\n<p>Good luck everybody anyway !</p>\n\n<p>Very interesting competition indeed.</p>\n\n<p>Hey guys my pizza is ready !!!</p>",
  "messages": [
    {
      "id": 315542,
      "postDate": "2018-04-17T08:57:54.777Z",
      "content": "<p>In that case stratified sampling might help ;)\nHowever you need to decide what 'classes' take into consideration during sampling.</p>",
      "rawMarkdown": "In that case stratified sampling might help ;)\nHowever you need to decide what 'classes' take into consideration during sampling.",
      "votes": 1
    },
    {
      "id": 312428,
      "postDate": "2018-04-11T18:13:37.023Z",
      "content": "<p>Be aware about those n:n rows that could lead to overfitting... Be somewhat generalizing, and be confident on your LB... That's my advice.</p>\n\n<p>Good luck everybody anyway !</p>\n\n<p>Very interesting competition indeed.</p>\n\n<p>Hey guys my pizza is ready !!!</p>",
      "rawMarkdown": "Be aware about those n:n rows that could lead to overfitting... Be somewhat generalizing, and be confident on your LB... That's my advice.\n\nGood luck everybody anyway !\n\nVery interesting competition indeed.\n\nHey guys my pizza is ready !!!",
      "votes": 1
    },
    {
      "id": 312622,
      "postDate": "2018-04-12T03:55:11.973Z",
      "rawMarkdown": "",
      "isDeleted": true,
      "replies": [
        {
          "id": 312635,
          "postDate": "2018-04-12T04:25:02.763Z",
          "content": "<p>Correct !</p>\n\n<p>Going to the end of file and leave the n first rows.</p>",
          "rawMarkdown": "Correct !\n\nGoing to the end of file and leave the n first rows.",
          "votes": 1
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 315542,
      "author_name": "Meyk",
      "author_url": "",
      "post_date": "2018-04-17T08:57:54.777000",
      "content": "<p>In that case stratified sampling might help ;)\nHowever you need to decide what 'classes' take into consideration during sampling.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 312622,
      "author_name": "",
      "author_url": "",
      "post_date": "2018-04-12T03:55:11.973000",
      "content": "",
      "votes": 0,
      "replies": [
        {
          "id": 312635,
          "author_name": "mezoganet",
          "author_url": "",
          "post_date": "2018-04-12T04:25:02.763000",
          "content": "<p>Correct !</p>\n\n<p>Going to the end of file and leave the n first rows.</p>",
          "votes": 1,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "315542": "In that case stratified sampling might help ;)\nHowever you need to decide what 'classes' take into consideration during sampling.",
    "312428": "Be aware about those n:n rows that could lead to overfitting... Be somewhat generalizing, and be confident on your LB... That's my advice.\n\nGood luck everybody anyway !\n\nVery interesting competition indeed.\n\nHey guys my pizza is ready !!!",
    "312622": ""
  }
}