{
  "id": 726,
  "title": "Goal is sorting?",
  "url": "/competitions/ClaimPredictionChallenge/discussion/726",
  "author_name": "",
  "post_date": "2011-07-24T14:59:30.860Z",
  "votes": null,
  "comment_count": 8,
  "views": 7554,
  "content": "<p>Anyone who can help: from what I understand it seems that the actual predicted claim dollar amount does not matter (contrary to what the competition data page suggests).&nbsp; Instead it seems as if the the goal is to sort the claims by amount (if I am understanding\r\n the GINI index correctly).&nbsp; Can anyone clarify?&nbsp; The GINI index seems only to care about ordering.</p>",
  "messages": [
    {
      "id": "4716",
      "postDate": "07/24/2011 14:59:30",
      "content": "<p>Anyone who can help: from what I understand it seems that the actual predicted claim dollar amount does not matter (contrary to what the competition data page suggests).&nbsp; Instead it seems as if the the goal is to sort the claims by amount (if I am understanding\r\n the GINI index correctly).&nbsp; Can anyone clarify?&nbsp; The GINI index seems only to care about ordering.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "4718",
      "postDate": "07/24/2011 15:21:10",
      "content": "<p>That's exactly right.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "4830",
      "postDate": "07/31/2011 22:12:45",
      "content": "<p>Insurer: do you mind clarifying your answer - the stated objective, and several posts suggest the aim is to predict the size of the claims - which is not the same as simply predicting relative claim order.</p>\r\n<p>Am also puzzled as to where the discussion of the Gini coefficient comes from.. I don't see any mention of it in this competition's description. I'm obviously missing something.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "4842",
      "postDate": "08/01/2011 17:15:08",
      "content": "<p>[quote=BVdS;4830]</p>\r\n<p>Insurer: do you mind clarifying your answer - the stated objective, and several posts suggest the aim is to predict the size of the claims - which is not the same as simply predicting relative claim order.</p>\r\n<p>[/quote]</p>\r\n<p>Only the relative order matters.The values don't mean anything on there own, it's just how they compare relative to each other.</p>\r\n<p>[quote=BVdS;4830]</p>\r\n<p>Am also puzzled as to where the discussion of the Gini coefficient comes from.. I don't see any mention of it in this competition's description. I'm obviously missing something.</p>\r\n<p>[/quote]</p>\r\n<p>The Gini coefficient only takes into account the relative order. Details about its calculation can be found at&nbsp;<a href=\"http://www.kaggle.com/c/ClaimPredictionChallenge/Details/Evaluation\">http://www.kaggle.com/c/ClaimPredictionChallenge/Details/Evaluation</a></p>\r\n<p>&nbsp;</p>\r\n<p>&nbsp;</p>\r\n<p>&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "4964",
      "postDate": "08/06/2011 11:30:15",
      "content": "<p>I am submitting only one column that is Row_id which is sorted descending on the basis of predicted claim amount. Am I on the right track</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "4965",
      "postDate": "08/06/2011 12:05:47",
      "content": "<p>[quote=Fractal;4964]</p>\r\n<p>I am submitting only one column that is Row_id which is sorted descending on the basis of predicted claim amount. Am I on the right track</p>\r\n<p>[/quote]</p>\r\n<p>The order of the rows in the submission must be sorted row_id ascending. The predicted values for each row_id must correspond to what you believe is the claim amount. We have to pair up your predictions with the actuals to do the scoring. For the exact details,\r\n see&nbsp;http://www.kaggle.com/c/ClaimPredictionChallenge/forums/t/703/code-to-calculate-normalizedgini . Note that we pair up actuals with predictions (as determined by row_id ascending sort) and then do a sort.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "4985",
      "postDate": "08/08/2011 02:36:31",
      "content": "<p>Hi Jeff,</p>\r\n<p>I've never come across&nbsp;Normalized GINI index performance evaluation term&nbsp;before.&nbsp;Since you pair up the actual&nbsp;values with the predicted values on ascending row_ids, then the key&nbsp;is to build a&nbsp;model which predicts claim_amount most accurately, therefore ranking\r\n would be better.&nbsp; Am I right?</p>\r\n<p>Regards,</p>\r\n<p>Seyhan</p>\r\n<p>&nbsp;</p>\r\n<p>&nbsp;</p>\r\n<p>&nbsp;&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "4986",
      "postDate": "08/08/2011 03:10:38",
      "content": "<p>Seyhan: as far as I understood predicted values are only used to determine the relative order of rows, so there's no need to actually predict claim_amount.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "5108",
      "postDate": "08/17/2011 21:02:47",
      "content": "<p>Normalized GINI simply takes into account the min and max of the particular scale of measure. As an economist, I use it when building wage disparity measures across countries. Since the currency systems are different, there is a need to standardize them.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 4718,
      "author_name": "allstateinsurancecompany",
      "author_url": "",
      "post_date": "07/24/2011 15:21:10",
      "content": "<p>That's exactly right.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 4830,
      "author_name": "bvds5431",
      "author_url": "",
      "post_date": "07/31/2011 22:12:45",
      "content": "<p>Insurer: do you mind clarifying your answer - the stated objective, and several posts suggest the aim is to predict the size of the claims - which is not the same as simply predicting relative claim order.</p>\r\n<p>Am also puzzled as to where the discussion of the Gini coefficient comes from.. I don't see any mention of it in this competition's description. I'm obviously missing something.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 4842,
      "author_name": "jeffmoser",
      "author_url": "",
      "post_date": "08/01/2011 17:15:08",
      "content": "<p>[quote=BVdS;4830]</p>\r\n<p>Insurer: do you mind clarifying your answer - the stated objective, and several posts suggest the aim is to predict the size of the claims - which is not the same as simply predicting relative claim order.</p>\r\n<p>[/quote]</p>\r\n<p>Only the relative order matters.The values don't mean anything on there own, it's just how they compare relative to each other.</p>\r\n<p>[quote=BVdS;4830]</p>\r\n<p>Am also puzzled as to where the discussion of the Gini coefficient comes from.. I don't see any mention of it in this competition's description. I'm obviously missing something.</p>\r\n<p>[/quote]</p>\r\n<p>The Gini coefficient only takes into account the relative order. Details about its calculation can be found at&nbsp;<a href=\"http://www.kaggle.com/c/ClaimPredictionChallenge/Details/Evaluation\">http://www.kaggle.com/c/ClaimPredictionChallenge/Details/Evaluation</a></p>\r\n<p>&nbsp;</p>\r\n<p>&nbsp;</p>\r\n<p>&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 4964,
      "author_name": "fractal",
      "author_url": "",
      "post_date": "08/06/2011 11:30:15",
      "content": "<p>I am submitting only one column that is Row_id which is sorted descending on the basis of predicted claim amount. Am I on the right track</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 4965,
      "author_name": "jeffmoser",
      "author_url": "",
      "post_date": "08/06/2011 12:05:47",
      "content": "<p>[quote=Fractal;4964]</p>\r\n<p>I am submitting only one column that is Row_id which is sorted descending on the basis of predicted claim amount. Am I on the right track</p>\r\n<p>[/quote]</p>\r\n<p>The order of the rows in the submission must be sorted row_id ascending. The predicted values for each row_id must correspond to what you believe is the claim amount. We have to pair up your predictions with the actuals to do the scoring. For the exact details,\r\n see&nbsp;http://www.kaggle.com/c/ClaimPredictionChallenge/forums/t/703/code-to-calculate-normalizedgini . Note that we pair up actuals with predictions (as determined by row_id ascending sort) and then do a sort.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 4985,
      "author_name": "seyhan",
      "author_url": "",
      "post_date": "08/08/2011 02:36:31",
      "content": "<p>Hi Jeff,</p>\r\n<p>I've never come across&nbsp;Normalized GINI index performance evaluation term&nbsp;before.&nbsp;Since you pair up the actual&nbsp;values with the predicted values on ascending row_ids, then the key&nbsp;is to build a&nbsp;model which predicts claim_amount most accurately, therefore ranking\r\n would be better.&nbsp; Am I right?</p>\r\n<p>Regards,</p>\r\n<p>Seyhan</p>\r\n<p>&nbsp;</p>\r\n<p>&nbsp;</p>\r\n<p>&nbsp;&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 4986,
      "author_name": "mikayelsukiasyan",
      "author_url": "",
      "post_date": "08/08/2011 03:10:38",
      "content": "<p>Seyhan: as far as I understood predicted values are only used to determine the relative order of rows, so there's no need to actually predict claim_amount.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 5108,
      "author_name": "ryerex",
      "author_url": "",
      "post_date": "08/17/2011 21:02:47",
      "content": "<p>Normalized GINI simply takes into account the min and max of the particular scale of measure. As an economist, I use it when building wage disparity measures across countries. Since the currency systems are different, there is a need to standardize them.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "4716": "",
    "4718": "",
    "4830": "",
    "4842": "",
    "4964": "",
    "4965": "",
    "4985": "",
    "4986": "",
    "5108": ""
  },
  "source": "meta"
}