{
  "id": 3056,
  "title": "new priors",
  "url": "/competitions/predict-closed-questions-on-stack-overflow/discussion/3056",
  "author_name": "",
  "post_date": "2012-11-02T18:12:54.797Z",
  "votes": null,
  "comment_count": 8,
  "views": 4656,
  "content": "<p>Perhaps anyone can tell what are prior probabilities in a newer training set? Old ones were: [0.00913477057600471, 0.004645859639795308, 0.005200965546050945, 0.9791913907850639, 0.0018270134530850952]. Too lazy to download whole file just to get these numbers.</p>\r\n<p>&nbsp;</p>\r\n<p>Thanks,</p>\r\n<p>ren</p>",
  "messages": [
    {
      "id": "16504",
      "postDate": "11/02/2012 18:12:54",
      "content": "<p>Perhaps anyone can tell what are prior probabilities in a newer training set? Old ones were: [0.00913477057600471, 0.004645859639795308, 0.005200965546050945, 0.9791913907850639, 0.0018270134530850952]. Too lazy to download whole file just to get these numbers.</p>\r\n<p>&nbsp;</p>\r\n<p>Thanks,</p>\r\n<p>ren</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "16505",
      "postDate": "11/02/2012 18:14:45",
      "content": "<p>#(0.010511190617039251d0 0.005701997568253705d0 0.0056883544674060866d0<br>\r\n0.9756672567762553d0 0.002431200571045629d0)</p>\r\n<p>But what matters though is the most recent data.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "16507",
      "postDate": "11/02/2012 18:20:59",
      "content": "<p>Thank you, that was quick:) Yes I read the previous post, perhaps you could also post priors for the last month only? Out of curiosity.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "16508",
      "postDate": "11/02/2012 18:25:34",
      "content": "<p>#(0.03410911204982466d0 0.01173872976800856d0 0.018430671606251586d0&nbsp;0.926642216133641d0 0.009079270442274271d0)</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "16512",
      "postDate": "11/02/2012 19:42:09",
      "content": "<p>There's been a real increase in # closed questions over the last few months (see graph). I guess (like this contest) it's all part of SO's initiative to keep questions on topic &amp; help new users ask appropriate questions etc.\r\n</p>\r\n<p>https://s3.amazonaws.com/fletcher/closed-questions-graph.png</p>\r\n<p>Makes the competition a bit harder if there's been big changes in the last few months though!</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "16513",
      "postDate": "11/02/2012 19:59:38",
      "content": "<p>The contests a bit harder, or the results more random.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "16531",
      "postDate": "11/03/2012 12:39:27",
      "content": "<p>yes, results are random.<br>\r\nSo if X used priors for the entire train set last time, going by here people can use different priors. Which means as long as algorithm is same, one can choose a different training set portion to train model?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "16537",
      "postDate": "11/03/2012 18:08:50",
      "content": "<p>Or you could have built a model which automatically used the priors from the most recent part of the data, after observing that the probabilities change with time ;)</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "16538",
      "postDate": "11/03/2012 18:47:56",
      "content": "",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 16505,
      "author_name": "melisgl",
      "author_url": "",
      "post_date": "11/02/2012 18:14:45",
      "content": "<p>#(0.010511190617039251d0 0.005701997568253705d0 0.0056883544674060866d0<br>\r\n0.9756672567762553d0 0.002431200571045629d0)</p>\r\n<p>But what matters though is the most recent data.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 16507,
      "author_name": "renz85",
      "author_url": "",
      "post_date": "11/02/2012 18:20:59",
      "content": "<p>Thank you, that was quick:) Yes I read the previous post, perhaps you could also post priors for the last month only? Out of curiosity.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 16508,
      "author_name": "melisgl",
      "author_url": "",
      "post_date": "11/02/2012 18:25:34",
      "content": "<p>#(0.03410911204982466d0 0.01173872976800856d0 0.018430671606251586d0&nbsp;0.926642216133641d0 0.009079270442274271d0)</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 16512,
      "author_name": "thomasfletcher",
      "author_url": "",
      "post_date": "11/02/2012 19:42:09",
      "content": "<p>There's been a real increase in # closed questions over the last few months (see graph). I guess (like this contest) it's all part of SO's initiative to keep questions on topic &amp; help new users ask appropriate questions etc.\r\n</p>\r\n<p>https://s3.amazonaws.com/fletcher/closed-questions-graph.png</p>\r\n<p>Makes the competition a bit harder if there's been big changes in the last few months though!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 16513,
      "author_name": "melisgl",
      "author_url": "",
      "post_date": "11/02/2012 19:59:38",
      "content": "<p>The contests a bit harder, or the results more random.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 16531,
      "author_name": "rkirana",
      "author_url": "",
      "post_date": "11/03/2012 12:39:27",
      "content": "<p>yes, results are random.<br>\r\nSo if X used priors for the entire train set last time, going by here people can use different priors. Which means as long as algorithm is same, one can choose a different training set portion to train model?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 16537,
      "author_name": "james30",
      "author_url": "",
      "post_date": "11/03/2012 18:08:50",
      "content": "<p>Or you could have built a model which automatically used the priors from the most recent part of the data, after observing that the probabilities change with time ;)</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 16538,
      "author_name": "rkirana",
      "author_url": "",
      "post_date": "11/03/2012 18:47:56",
      "content": "",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "16504": "",
    "16505": "",
    "16507": "",
    "16508": "",
    "16512": "",
    "16513": "",
    "16531": "",
    "16537": "",
    "16538": ""
  },
  "source": "meta"
}