{
  "id": 52721,
  "title": "Why using label encoding on channel, app, device and os ?",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/52721",
  "author_name": "maxpv",
  "post_date": "2018-03-22T13:25:33.384000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Is it a problem if a feature is not continuous (example: [0,1,2,3,5,6], ie no 4) ?</p>\n\n<p>I see a lot of kernels which use label encoding for those features.</p>",
  "messages": [
    {
      "id": 301203,
      "postDate": "2018-03-22T13:25:33.383Z",
      "content": "<p>Is it a problem if a feature is not continuous (example: [0,1,2,3,5,6], ie no 4) ?</p>\n\n<p>I see a lot of kernels which use label encoding for those features.</p>",
      "rawMarkdown": "Is it a problem if a feature is not continuous (example: [0,1,2,3,5,6], ie no 4) ?\n\nI see a lot of kernels which use label encoding for those features.",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "301203": "Is it a problem if a feature is not continuous (example: [0,1,2,3,5,6], ie no 4) ?\n\nI see a lot of kernels which use label encoding for those features."
  }
}