{
  "id": 16480,
  "title": "How many sponsored htmls left for us to classify?",
  "url": "/competitions/dato-native/discussion/16480",
  "author_name": "",
  "post_date": "2015-09-14T15:15:11.653Z",
  "votes": null,
  "comment_count": 2,
  "views": 483,
  "content": "<p>Can we estimate how many sponsored htmls left for us to classify from the score 0.99805, I have the rough idea of how ROC is calculated, but not sure if we can infer how many sponsored htmls are left from it ?</p>\n\n<p>If too few left, then it is more or less like gamble, I am wondering if the admin have more test data that can be released for us to classify ? </p>",
  "messages": [
    {
      "id": "92493",
      "postDate": "09/14/2015 15:15:11",
      "content": "<p>Can we estimate how many sponsored htmls left for us to classify from the score 0.99805, I have the rough idea of how ROC is calculated, but not sure if we can infer how many sponsored htmls are left from it ?</p>\n\n<p>If too few left, then it is more or less like gamble, I am wondering if the admin have more test data that can be released for us to classify ? </p>",
      "rawMarkdown": "Can we estimate how many sponsored htmls left for us to classify from the score 0.99805, I have the rough idea of how ROC is calculated, but not sure if we can infer how many sponsored htmls are left from it ?\r\n\r\nIf too few left, then it is more or less like gamble, I am wondering if the admin have more test data that can be released for us to classify ?",
      "votes": null
    },
    {
      "id": "92494",
      "postDate": "09/14/2015 15:28:58",
      "content": "<p>less than 1%. The area under the curve is basically a  square. Since the area is 0.99805 =&gt; the length at fpr axis &gt; = 0.99805 and the length at tpr axis &gt;=0.99805. So we have less than 1% sponsored htmls left.</p>\n\n<p>Is it right?</p>",
      "rawMarkdown": "less than 1%. The area under the curve is basically a  square. Since the area is 0.99805 => the length at fpr axis > = 0.99805 and the length at tpr axis >=0.99805. So we have less than 1% sponsored htmls left.\r\n\r\nIs it right?",
      "votes": null
    },
    {
      "id": "92504",
      "postDate": "09/14/2015 16:05:24",
      "content": "<p>&quot;Guess what? This competition has a fever and the only prescription is more test data.&quot; :)</p>\n\n<p>I am pretty sure StumbleUpon has a lot more data where the original datasets came from - I bet they are working on preparing more test data as we speak.</p>",
      "rawMarkdown": "\"Guess what? This competition has a fever and the only prescription is more test data.\" :)\r\n\r\nI am pretty sure StumbleUpon has a lot more data where the original datasets came from - I bet they are working on preparing more test data as we speak.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 92494,
      "author_name": "jiweiliu",
      "author_url": "",
      "post_date": "09/14/2015 15:28:58",
      "content": "<p>less than 1%. The area under the curve is basically a  square. Since the area is 0.99805 =&gt; the length at fpr axis &gt; = 0.99805 and the length at tpr axis &gt;=0.99805. So we have less than 1% sponsored htmls left.</p>\n\n<p>Is it right?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 92504,
      "author_name": "bekbolatov",
      "author_url": "",
      "post_date": "09/14/2015 16:05:24",
      "content": "<p>&quot;Guess what? This competition has a fever and the only prescription is more test data.&quot; :)</p>\n\n<p>I am pretty sure StumbleUpon has a lot more data where the original datasets came from - I bet they are working on preparing more test data as we speak.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "92493": "Can we estimate how many sponsored htmls left for us to classify from the score 0.99805, I have the rough idea of how ROC is calculated, but not sure if we can infer how many sponsored htmls are left from it ?\r\n\r\nIf too few left, then it is more or less like gamble, I am wondering if the admin have more test data that can be released for us to classify ?",
    "92494": "less than 1%. The area under the curve is basically a  square. Since the area is 0.99805 => the length at fpr axis > = 0.99805 and the length at tpr axis >=0.99805. So we have less than 1% sponsored htmls left.\r\n\r\nIs it right?",
    "92504": "\"Guess what? This competition has a fever and the only prescription is more test data.\" :)\r\n\r\nI am pretty sure StumbleUpon has a lot more data where the original datasets came from - I bet they are working on preparing more test data as we speak."
  },
  "source": "meta"
}