{
  "id": 4718,
  "title": "Any update on test data and validation set?",
  "url": "/competitions/cause-effect-pairs/discussion/4718",
  "author_name": "",
  "post_date": "2013-05-31T08:41:51.013Z",
  "votes": null,
  "comment_count": 3,
  "views": 1594,
  "content": "<p>Any update?</p>\r\n<p></p>",
  "messages": [
    {
      "id": "24978",
      "postDate": "05/31/2013 08:41:51",
      "content": "<p>Any update?</p>\r\n<p></p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "25114",
      "postDate": "06/03/2013 23:15:23",
      "content": "<p>Kaggle has agreed that we would replace the validation set. When the change happens, all scores will be manually decreased by 0.5 and the participants will be invited to make a new submission to update their score.<br>\r\n<br>\r\nWe are presently performing further checks on the new artificial data and collecting more real data to prepare the final evaluation set. After following the debate on artificial vs. real data, we still intend to include both to mitigate the pros and cons.</p>\r\n<p>When the data are ready, a subset of them will compose the new vaildation set and the change will be made. At the same time, the test data will be made available for download encrypted. The decryption key will be made available only on July 19 (a few days\r\n after the software submission deadline of July 15).</p>\r\n<p>We intend to make the new validation data and the (encrypted) test data available as soon as possible and no later than July 1, which will give to the participants at least 2 weeks to download the test data and make any adjustment necessary to their software.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "25291",
      "postDate": "06/08/2013 20:12:19",
      "content": "<p>Great news. Will there be an additional column saying whether or not the data is artificial?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "25294",
      "postDate": "06/08/2013 21:34:56",
      "content": "<p>I don't think so. I think it is preferable that the participants develop a unified method rather that try to exploit specific features of artificial or real data.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 25114,
      "author_name": "iguyon",
      "author_url": "",
      "post_date": "06/03/2013 23:15:23",
      "content": "<p>Kaggle has agreed that we would replace the validation set. When the change happens, all scores will be manually decreased by 0.5 and the participants will be invited to make a new submission to update their score.<br>\r\n<br>\r\nWe are presently performing further checks on the new artificial data and collecting more real data to prepare the final evaluation set. After following the debate on artificial vs. real data, we still intend to include both to mitigate the pros and cons.</p>\r\n<p>When the data are ready, a subset of them will compose the new vaildation set and the change will be made. At the same time, the test data will be made available for download encrypted. The decryption key will be made available only on July 19 (a few days\r\n after the software submission deadline of July 15).</p>\r\n<p>We intend to make the new validation data and the (encrypted) test data available as soon as possible and no later than July 1, which will give to the participants at least 2 weeks to download the test data and make any adjustment necessary to their software.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 25291,
      "author_name": "ahassaine",
      "author_url": "",
      "post_date": "06/08/2013 20:12:19",
      "content": "<p>Great news. Will there be an additional column saying whether or not the data is artificial?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 25294,
      "author_name": "iguyon",
      "author_url": "",
      "post_date": "06/08/2013 21:34:56",
      "content": "<p>I don't think so. I think it is preferable that the participants develop a unified method rather that try to exploit specific features of artificial or real data.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "24978": "",
    "25114": "",
    "25291": "",
    "25294": ""
  },
  "source": "meta"
}