{
  "id": 120303,
  "title": "Multi-Model Solution",
  "url": "/competitions/ashrae-energy-prediction/discussion/120303",
  "author_name": "Aniket Goel",
  "post_date": "2019-12-05T04:45:44.210000",
  "votes": 0,
  "comment_count": 2,
  "views": 0,
  "content": "<p>If we train multiple models per building &amp; meter type then metadata of building or complications of different meter type will not play any role.</p>\n\n<p>Did anyone try this route?</p>\n\n<p>Only variables that would be eligible would then be weather information (which itself is very dirty but anyways needs to be handled in any case).</p>",
  "messages": [
    {
      "id": 688540,
      "postDate": "2019-12-05T18:01:46.083Z",
      "content": "<p>This direction of finding a solution of course must be investigated, the data is very noisy, so can win a simple solution with blending, which successfully minimizes noise and line up through the middle ground. Good luck!</p>",
      "rawMarkdown": "This direction of finding a solution of course must be investigated, the data is very noisy, so can win a simple solution with blending, which successfully minimizes noise and line up through the middle ground. Good luck!",
      "votes": 1
    },
    {
      "id": 687975,
      "postDate": "2019-12-05T04:45:44.210Z",
      "content": "<p>If we train multiple models per building &amp; meter type then metadata of building or complications of different meter type will not play any role.</p>\n\n<p>Did anyone try this route?</p>\n\n<p>Only variables that would be eligible would then be weather information (which itself is very dirty but anyways needs to be handled in any case).</p>",
      "rawMarkdown": "If we train multiple models per building &amp; meter type then metadata of building or complications of different meter type will not play any role.\n\nDid anyone try this route?\n\nOnly variables that would be eligible would then be weather information (which itself is very dirty but anyways needs to be handled in any case)."
    },
    {
      "id": 690739,
      "postDate": "2019-12-09T03:08:02.260Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 688540,
      "author_name": "Aleksandr Kosolapov",
      "author_url": "",
      "post_date": "2019-12-05T18:01:46.083000",
      "content": "<p>This direction of finding a solution of course must be investigated, the data is very noisy, so can win a simple solution with blending, which successfully minimizes noise and line up through the middle ground. Good luck!</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 690739,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-12-09T03:08:02.260000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "688540": "This direction of finding a solution of course must be investigated, the data is very noisy, so can win a simple solution with blending, which successfully minimizes noise and line up through the middle ground. Good luck!",
    "687975": "If we train multiple models per building &amp; meter type then metadata of building or complications of different meter type will not play any role.\n\nDid anyone try this route?\n\nOnly variables that would be eligible would then be weather information (which itself is very dirty but anyways needs to be handled in any case).",
    "690739": ""
  }
}