{
  "id": 3020,
  "title": "Making Your Final Submission",
  "url": "/competitions/predict-closed-questions-on-stack-overflow/discussion/3020",
  "author_name": "",
  "post_date": "2012-10-29T21:21:50.260Z",
  "votes": null,
  "comment_count": 1,
  "views": 1908,
  "content": "<p>Hi everyone,</p>\r\n<p>As a reminder, this competition is closing in 5 days. In order to be eligible for prize and receive a final rank for this competition, you need to make a submission on the private leaderboard set. This can be downloaded as private_leaderboard.csv from\r\n<a href=\"https://www.kaggle.com/c/predict-closed-questions-on-stack-overflow/data\">\r\ndata page</a>.</p>\r\n<p>Additionally, you have the option of retraining the models you submitted on an updated training set (which contains data through October 9, 2012). These may be downloaded from the\r\n<a href=\"https://www.kaggle.com/c/predict-closed-questions-on-stack-overflow/data\">\r\ndata page</a> as train_October_9_2012.csv (for the full version) or train-sample_October_9_2012.csv (for the stratified sample).</p>\r\n<p>When you make a final submission, it will show up briefly on the leaderboard with a score of 0.0000 (visible only to you). This means it has been parsed correctly, and the private leaderboard score will be unveiled at the end of the competition.</p>\r\n<p>Also, the<a href=\"https://www.kaggle.com/c/predict-closed-questions-on-stack-overflow/prospector\"> visualization contest</a> has now closed. There have been many outstanding entries - vote for your favorites!</p>\r\n<p>Thanks to all for your hard work on this contest.</p>\r\n<p>Ben</p>",
  "messages": [
    {
      "id": "16344",
      "postDate": "10/29/2012 21:21:50",
      "content": "<p>Hi everyone,</p>\r\n<p>As a reminder, this competition is closing in 5 days. In order to be eligible for prize and receive a final rank for this competition, you need to make a submission on the private leaderboard set. This can be downloaded as private_leaderboard.csv from\r\n<a href=\"https://www.kaggle.com/c/predict-closed-questions-on-stack-overflow/data\">\r\ndata page</a>.</p>\r\n<p>Additionally, you have the option of retraining the models you submitted on an updated training set (which contains data through October 9, 2012). These may be downloaded from the\r\n<a href=\"https://www.kaggle.com/c/predict-closed-questions-on-stack-overflow/data\">\r\ndata page</a> as train_October_9_2012.csv (for the full version) or train-sample_October_9_2012.csv (for the stratified sample).</p>\r\n<p>When you make a final submission, it will show up briefly on the leaderboard with a score of 0.0000 (visible only to you). This means it has been parsed correctly, and the private leaderboard score will be unveiled at the end of the competition.</p>\r\n<p>Also, the<a href=\"https://www.kaggle.com/c/predict-closed-questions-on-stack-overflow/prospector\"> visualization contest</a> has now closed. There have been many outstanding entries - vote for your favorites!</p>\r\n<p>Thanks to all for your hard work on this contest.</p>\r\n<p>Ben</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "16352",
      "postDate": "10/30/2012 06:54:05",
      "content": "<p>Hi Ben,</p>\r\n<p>Although the current feedback (&quot;a score of 0.0000&quot;) &nbsp;can tell us whether the submission format is correct or not, we may still make other mistakes, that have not been encountered in the first phrase. Thus, could you publish the scores of the benchmarks on\r\n private leaderboard or provide the true feedback, the real log-loss of a submission, in limited quota, e.g., one chance per team? &nbsp;</p>\r\n<p>These can definitely help us build the submission confidently.&nbsp; Thanks.</p>\r\n<p>&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 16352,
      "author_name": "",
      "author_url": "",
      "post_date": "10/30/2012 06:54:05",
      "content": "<p>Hi Ben,</p>\r\n<p>Although the current feedback (&quot;a score of 0.0000&quot;) &nbsp;can tell us whether the submission format is correct or not, we may still make other mistakes, that have not been encountered in the first phrase. Thus, could you publish the scores of the benchmarks on\r\n private leaderboard or provide the true feedback, the real log-loss of a submission, in limited quota, e.g., one chance per team? &nbsp;</p>\r\n<p>These can definitely help us build the submission confidently.&nbsp; Thanks.</p>\r\n<p>&nbsp;</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "16344": "",
    "16352": ""
  },
  "source": "meta"
}