{
  "id": 20865,
  "title": "How do you do feature selection here?",
  "url": "/competitions/expedia-hotel-recommendations/discussion/20865",
  "author_name": "",
  "post_date": "2016-05-11T14:02:47.190Z",
  "votes": null,
  "comment_count": 2,
  "views": 807,
  "content": "<p>I thought of trying to use random forest for that, But not enough memory for the one hot encoding.\nHow do you do feature selection?</p>",
  "messages": [
    {
      "id": "119577",
      "postDate": "05/11/2016 14:02:47",
      "content": "<p>I thought of trying to use random forest for that, But not enough memory for the one hot encoding.\nHow do you do feature selection?</p>",
      "rawMarkdown": "I thought of trying to use random forest for that, But not enough memory for the one hot encoding.\r\nHow do you do feature selection?",
      "votes": null
    },
    {
      "id": "119618",
      "postDate": "05/11/2016 22:17:08",
      "content": "<p>I think most people here is trying to down sample or clustering the data set into partitions. Then one hot encoding might be easier.</p>",
      "rawMarkdown": "I think most people here is trying to down sample or clustering the data set into partitions. Then one hot encoding might be easier.",
      "votes": null
    },
    {
      "id": "121483",
      "postDate": "05/26/2016 20:21:50",
      "content": "<p>Why do you need  to do one-hot encoding here?  I learned that one-hot encoding is not useful for tree classifier? Thank you.</p>",
      "rawMarkdown": "Why do you need  to do one-hot encoding here?  I learned that one-hot encoding is not useful for tree classifier? Thank you.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 119618,
      "author_name": "aikinogard",
      "author_url": "",
      "post_date": "05/11/2016 22:17:08",
      "content": "<p>I think most people here is trying to down sample or clustering the data set into partitions. Then one hot encoding might be easier.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 121483,
      "author_name": "beedata",
      "author_url": "",
      "post_date": "05/26/2016 20:21:50",
      "content": "<p>Why do you need  to do one-hot encoding here?  I learned that one-hot encoding is not useful for tree classifier? Thank you.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "119577": "I thought of trying to use random forest for that, But not enough memory for the one hot encoding.\r\nHow do you do feature selection?",
    "119618": "I think most people here is trying to down sample or clustering the data set into partitions. Then one hot encoding might be easier.",
    "121483": "Why do you need  to do one-hot encoding here?  I learned that one-hot encoding is not useful for tree classifier? Thank you."
  },
  "source": "meta"
}