{
  "id": 27375,
  "title": "Namespace in FFM Feature Hasing",
  "url": "/competitions/outbrain-click-prediction/discussion/27375",
  "author_name": "Nhuan Tran",
  "post_date": "2017-01-06T03:12:58.907000",
  "votes": 0,
  "comment_count": 0,
  "views": 173,
  "content": "<p>Preparing FFM Features, I used murmush32 to hash namespace with value, for example ad_id_1234 =&gt; 83827101. I wonder length of namespace will affect to the model? \nBecause when I pick namespace 'd' instead of 'ad_id', my score reaches to public 0.63. On the other hands, i pick namespace 'ad_id', the results is terrible wrong. \nCan someone explain me why?</p>",
  "messages": [
    {
      "id": 154409,
      "postDate": "2017-01-06T03:12:58.907Z",
      "content": "<p>Preparing FFM Features, I used murmush32 to hash namespace with value, for example ad_id_1234 =&gt; 83827101. I wonder length of namespace will affect to the model? \nBecause when I pick namespace 'd' instead of 'ad_id', my score reaches to public 0.63. On the other hands, i pick namespace 'ad_id', the results is terrible wrong. \nCan someone explain me why?</p>",
      "rawMarkdown": "Preparing FFM Features, I used murmush32 to hash namespace with value, for example ad_id_1234 => 83827101. I wonder length of namespace will affect to the model? \r\nBecause when I pick namespace 'd' instead of 'ad_id', my score reaches to public 0.63. On the other hands, i pick namespace 'ad_id', the results is terrible wrong. \r\nCan someone explain me why?"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "154409": "Preparing FFM Features, I used murmush32 to hash namespace with value, for example ad_id_1234 => 83827101. I wonder length of namespace will affect to the model? \r\nBecause when I pick namespace 'd' instead of 'ad_id', my score reaches to public 0.63. On the other hands, i pick namespace 'ad_id', the results is terrible wrong. \r\nCan someone explain me why?"
  }
}