{
  "id": 8105,
  "title": "Do you treat this problem as a classfication problem and use models such as random forest and gbdt?",
  "url": "/competitions/acquire-valued-shoppers-challenge/discussion/8105",
  "author_name": "",
  "post_date": "2014-05-10T03:32:32.163Z",
  "votes": null,
  "comment_count": 1,
  "views": 1055,
  "content": "<p>I have not extract enough features and got a bad performance by using random forest.Maybe I could do better when I extract more features.</p>",
  "messages": [
    {
      "id": "44328",
      "postDate": "05/10/2014 03:32:32",
      "content": "<p>I have not extract enough features and got a bad performance by using random forest.Maybe I could do better when I extract more features.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "44370",
      "postDate": "05/11/2014 11:27:16",
      "content": "<p>I think you need to clarify what's exactly the problem.</p>\n<p>As&nbsp;a starting point you can use counts&nbsp;of transaction per same (brand - category - company) triad as an offer for&nbsp;different time periods. (One week, Two weeks, One month, Three month ...)</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 44370,
      "author_name": "dremovd",
      "author_url": "",
      "post_date": "05/11/2014 11:27:16",
      "content": "<p>I think you need to clarify what's exactly the problem.</p>\n<p>As&nbsp;a starting point you can use counts&nbsp;of transaction per same (brand - category - company) triad as an offer for&nbsp;different time periods. (One week, Two weeks, One month, Three month ...)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "44328": "",
    "44370": ""
  },
  "source": "meta"
}