{
  "id": 117436,
  "title": "Building Id as category Vs Building Id as Numeric Boosts Lb?",
  "url": "/competitions/ashrae-energy-prediction/discussion/117436",
  "author_name": "AdityaVikramSingh",
  "post_date": "2019-11-15T14:42:46.158000",
  "votes": 5,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi , I noticed something strange for my models , I changes a very variables(like - Building id &amp; Site id etc ) from categorical to numerical  and when i submitted i got a jump on the leaderboard , does anyone knowwhat could be the difference in treating a variable as categorical vs Numerical for tree based models (XGB or LGB ) . I would have understood the difference if there were some categories present only in the test set but it's not the case here at all</p>",
  "messages": [
    {
      "id": 673839,
      "postDate": "2019-11-15T14:42:46.160Z",
      "content": "<p>Hi , I noticed something strange for my models , I changes a very variables(like - Building id &amp; Site id etc ) from categorical to numerical  and when i submitted i got a jump on the leaderboard , does anyone knowwhat could be the difference in treating a variable as categorical vs Numerical for tree based models (XGB or LGB ) . I would have understood the difference if there were some categories present only in the test set but it's not the case here at all</p>",
      "rawMarkdown": "Hi , I noticed something strange for my models , I changes a very variables(like - Building id &amp; Site id etc ) from categorical to numerical  and when i submitted i got a jump on the leaderboard , does anyone knowwhat could be the difference in treating a variable as categorical vs Numerical for tree based models (XGB or LGB ) . I would have understood the difference if there were some categories present only in the test set but it's not the case here at all",
      "votes": 5
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "673839": "Hi , I noticed something strange for my models , I changes a very variables(like - Building id &amp; Site id etc ) from categorical to numerical  and when i submitted i got a jump on the leaderboard , does anyone knowwhat could be the difference in treating a variable as categorical vs Numerical for tree based models (XGB or LGB ) . I would have understood the difference if there were some categories present only in the test set but it's not the case here at all"
  }
}