{
  "id": 10749,
  "title": "Log loss evaluation metric",
  "url": "/competitions/avazu-ctr-prediction/discussion/10749",
  "author_name": "",
  "post_date": "2014-10-25T22:31:41.563Z",
  "votes": null,
  "comment_count": 1,
  "views": 2123,
  "content": "<p>For evaluation, is log loss an average of log loss over every line?</p>\n<p>In my code I have:</p>\n<p style=\"padding-left: 30px\">for row</p>\n<p style=\"padding-left: 60px\">loss += logloss(p,y)</p>\n<p style=\"padding-left: 60px\">line += 1&nbsp;</p>\n\n<p>Should I evaluate myself on loss / line, or just loss?</p>",
  "messages": [
    {
      "id": "56773",
      "postDate": "10/25/2014 22:31:41",
      "content": "<p>For evaluation, is log loss an average of log loss over every line?</p>\n<p>In my code I have:</p>\n<p style=\"padding-left: 30px\">for row</p>\n<p style=\"padding-left: 60px\">loss += logloss(p,y)</p>\n<p style=\"padding-left: 60px\">line += 1&nbsp;</p>\n\n<p>Should I evaluate myself on loss / line, or just loss?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "56775",
      "postDate": "10/25/2014 23:31:13",
      "content": "<p>loss/line, please refer to&nbsp;https://www.kaggle.com/wiki/LogarithmicLoss</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 56775,
      "author_name": "tandyzhu",
      "author_url": "",
      "post_date": "10/25/2014 23:31:13",
      "content": "<p>loss/line, please refer to&nbsp;https://www.kaggle.com/wiki/LogarithmicLoss</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "56773": "",
    "56775": ""
  },
  "source": "meta"
}