{
  "id": 3658,
  "title": "A question About Timeline",
  "url": "/competitions/flight/discussion/3658",
  "author_name": "",
  "post_date": "2013-01-22T01:27:48.923Z",
  "votes": 1,
  "comment_count": 6,
  "views": 4292,
  "content": "<p>Hi Kaggle Admins/Members,</p>\r\n<p>&nbsp;</p>\r\n<p>I see a contradiction between timeline/submission instructions. Based on timeline:</p>\r\n<p><strong>February 14, 2013 :&nbsp;Model Submission Deadline (must include code or hash of code)</strong></p>\r\n<p><strong>February 8, 2013</strong><br>\r\n<strong>Augmented Training Set, Part 2 Released</strong><br>\r\n<strong>Public Leaderboard is frozen and Public Leaderboard test data (with ground truth arrival times) is released</strong></p>\r\n<p><span>Based on submission instruction:</span></p>\r\n<p><strong>In order to be eligible for prizes, you must make at least one submission for the public leaderboard set using the same model you will use for the final evaluation set.</strong></p>\r\n<p>When is the deadline for model? February 8th or 14th? Can we still improve the model between Feb 8th and 14th?</p>\r\n<p><span style=\"line-height:1.4em\">Thanks&nbsp;</span></p>\r\n<p>&nbsp;</p>\r\n<p><span>&nbsp;</span></p>",
  "messages": [
    {
      "id": "19573",
      "postDate": "01/22/2013 01:27:48",
      "content": "<p>Hi Kaggle Admins/Members,</p>\r\n<p>&nbsp;</p>\r\n<p>I see a contradiction between timeline/submission instructions. Based on timeline:</p>\r\n<p><strong>February 14, 2013 :&nbsp;Model Submission Deadline (must include code or hash of code)</strong></p>\r\n<p><strong>February 8, 2013</strong><br>\r\n<strong>Augmented Training Set, Part 2 Released</strong><br>\r\n<strong>Public Leaderboard is frozen and Public Leaderboard test data (with ground truth arrival times) is released</strong></p>\r\n<p><span>Based on submission instruction:</span></p>\r\n<p><strong>In order to be eligible for prizes, you must make at least one submission for the public leaderboard set using the same model you will use for the final evaluation set.</strong></p>\r\n<p>When is the deadline for model? February 8th or 14th? Can we still improve the model between Feb 8th and 14th?</p>\r\n<p><span style=\"line-height:1.4em\">Thanks&nbsp;</span></p>\r\n<p>&nbsp;</p>\r\n<p><span>&nbsp;</span></p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "19838",
      "postDate": "01/30/2013 02:39:49",
      "content": "<p>Bumping this post.</p>\r\n<p>&nbsp;</p>\r\n<p>We are now about a week away from the 8th deadline and it seems like this is a really important clarification.&nbsp; Can we change our model between the 8th and the 14th?<br>\r\n<br>\r\n<em>Entrants further develop models using all available data and choose their best model as a final model. No further entries are submitted to the leaderboard.</em><br>\r\n<br>\r\nCan someone define &quot;further develop models&quot;?&nbsp; Does this mean a re-training on current models with new data? Does this mean as long as we have submitted one score to public LB we can continue developing models?&nbsp; Does this mean we are free to choose from say\r\n 1 of several models as long as each model has submitted an entry to the public LB?<br>\r\n<br>\r\nI think everyone here who has worked hard in this competition would hate to get disqualified because of any ambiguity here.<br>\r\n<br>\r\nThanks in advance.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "19840",
      "postDate": "01/30/2013 03:17:43",
      "content": "<p>[quote=ATCGuy;19838]</p>\r\n<p>Bumping this post.</p>\r\n<p>&nbsp;</p>\r\n<p>We are now about a week away from the 8th deadline and it seems like this is a really important clarification.&nbsp; Can we change our model between the 8th and the 14th?<br>\r\n<br>\r\n<em>Entrants further develop models using all available data and choose their best model as a final model. No further entries are submitted to the leaderboard.</em><br>\r\n<br>\r\nCan someone define &quot;further develop models&quot;?&nbsp; Does this mean a re-training on current models with new data? Does this mean as long as we have submitted one score to public LB we can continue developing models?&nbsp; Does this mean we are free to choose from say\r\n 1 of several models as long as each model has submitted an entry to the public LB?<br>\r\n<br>\r\nI think everyone here who has worked hard in this competition would hate to get disqualified because of any ambiguity here.<br>\r\n<br>\r\nThanks in advance.</p>\r\n<p>[/quote]Apologies for any ambiguity in the rules. You need to have at least one submission to the public LB prior to the Feb 8th deadline. You may retrain and update your model based on the additional data that we release on Feb 8/9th. We'll send out detailed\r\n instructions on this process a week before the Feb 8th deadline.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "19948",
      "postDate": "02/02/2013 14:39:19",
      "content": "<p>We are now within one week of Feb 8 and &quot;Retrain and update&quot; doesn't really make it clearer. So what am I allowed to do between Feb 8 and 14?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "19953",
      "postDate": "02/03/2013 00:03:07",
      "content": "As long as you have made at least one submission to the public leader board you can continue developing models.",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "19964",
      "postDate": "02/03/2013 08:42:58",
      "content": "<p>Thanks. And the submission could be from an entirely different algorithm? If so, it's a strange restriction.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "19966",
      "postDate": "02/03/2013 08:55:44",
      "content": "<p>[quote=Gábor Melis;19964]</p>\r\n<p>Thanks. And the submission could be from an entirely different algorithm? If so, it's a strange restriction.</p>\r\n<p>[/quote]It's a technical restriction on our platform (you won't be able to make additional submissions to the public leaderboard after February 8th, and you need an existing submission to attach the model code / hash to). Plus, we wanted to make sure everyone\r\n entering the competition made at least one submission to the public leaderboard.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 19838,
      "author_name": "mikeofalltrades",
      "author_url": "",
      "post_date": "01/30/2013 02:39:49",
      "content": "<p>Bumping this post.</p>\r\n<p>&nbsp;</p>\r\n<p>We are now about a week away from the 8th deadline and it seems like this is a really important clarification.&nbsp; Can we change our model between the 8th and the 14th?<br>\r\n<br>\r\n<em>Entrants further develop models using all available data and choose their best model as a final model. No further entries are submitted to the leaderboard.</em><br>\r\n<br>\r\nCan someone define &quot;further develop models&quot;?&nbsp; Does this mean a re-training on current models with new data? Does this mean as long as we have submitted one score to public LB we can continue developing models?&nbsp; Does this mean we are free to choose from say\r\n 1 of several models as long as each model has submitted an entry to the public LB?<br>\r\n<br>\r\nI think everyone here who has worked hard in this competition would hate to get disqualified because of any ambiguity here.<br>\r\n<br>\r\nThanks in advance.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 19840,
      "author_name": "benhamner",
      "author_url": "",
      "post_date": "01/30/2013 03:17:43",
      "content": "<p>[quote=ATCGuy;19838]</p>\r\n<p>Bumping this post.</p>\r\n<p>&nbsp;</p>\r\n<p>We are now about a week away from the 8th deadline and it seems like this is a really important clarification.&nbsp; Can we change our model between the 8th and the 14th?<br>\r\n<br>\r\n<em>Entrants further develop models using all available data and choose their best model as a final model. No further entries are submitted to the leaderboard.</em><br>\r\n<br>\r\nCan someone define &quot;further develop models&quot;?&nbsp; Does this mean a re-training on current models with new data? Does this mean as long as we have submitted one score to public LB we can continue developing models?&nbsp; Does this mean we are free to choose from say\r\n 1 of several models as long as each model has submitted an entry to the public LB?<br>\r\n<br>\r\nI think everyone here who has worked hard in this competition would hate to get disqualified because of any ambiguity here.<br>\r\n<br>\r\nThanks in advance.</p>\r\n<p>[/quote]Apologies for any ambiguity in the rules. You need to have at least one submission to the public LB prior to the Feb 8th deadline. You may retrain and update your model based on the additional data that we release on Feb 8/9th. We'll send out detailed\r\n instructions on this process a week before the Feb 8th deadline.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 19948,
      "author_name": "melisgl",
      "author_url": "",
      "post_date": "02/02/2013 14:39:19",
      "content": "<p>We are now within one week of Feb 8 and &quot;Retrain and update&quot; doesn't really make it clearer. So what am I allowed to do between Feb 8 and 14?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 19953,
      "author_name": "benhamner",
      "author_url": "",
      "post_date": "02/03/2013 00:03:07",
      "content": "As long as you have made at least one submission to the public leader board you can continue developing models.",
      "votes": null,
      "replies": []
    },
    {
      "id": 19964,
      "author_name": "melisgl",
      "author_url": "",
      "post_date": "02/03/2013 08:42:58",
      "content": "<p>Thanks. And the submission could be from an entirely different algorithm? If so, it's a strange restriction.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 19966,
      "author_name": "benhamner",
      "author_url": "",
      "post_date": "02/03/2013 08:55:44",
      "content": "<p>[quote=Gábor Melis;19964]</p>\r\n<p>Thanks. And the submission could be from an entirely different algorithm? If so, it's a strange restriction.</p>\r\n<p>[/quote]It's a technical restriction on our platform (you won't be able to make additional submissions to the public leaderboard after February 8th, and you need an existing submission to attach the model code / hash to). Plus, we wanted to make sure everyone\r\n entering the competition made at least one submission to the public leaderboard.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "19573": "",
    "19838": "",
    "19840": "",
    "19948": "",
    "19953": "",
    "19964": "",
    "19966": ""
  },
  "source": "meta"
}