{
  "id": 51900,
  "title": "I have a guess that the price of user's phone may play a role in predicting?",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/51900",
  "author_name": "",
  "post_date": "2018-03-14T06:25:32.829999700Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Is there any map to use for decoding the phone codec in data?\nThank you!</p>",
  "messages": [
    {
      "id": "295762",
      "postDate": "03/14/2018 06:25:32",
      "content": "<p>Is there any map to use for decoding the phone codec in data?\nThank you!</p>",
      "rawMarkdown": "Is there any map to use for decoding the phone codec in data?\nThank you!",
      "votes": null
    },
    {
      "id": "301785",
      "postDate": "03/23/2018 08:44:36",
      "content": "<p>I wondered the same, but unmasking <strong><em><code>device</code></em></strong> seems to be futile, hence <strong><em><code>os</code></em></strong> will mix variants of both OSes, phones. Let us continue the discussion in the thread <a href=\"https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/52201\">\"Freq counts from Train &amp; Test datasets\"</a>, see his spreadsheet.</p>\n\n<p>(FYI, given this is from China in Nov 2017, just before Singles Day, the best prior estimate for smartphone distribution might be \"top phones in China 2017\" or else \"2016\". But anyway that seems useless if <strong><em><code>os</code></em></strong> combines device and Android fragmentation. Anyway let's continue all this in the other thread.)</p>",
      "rawMarkdown": "I wondered the same, but unmasking ***`device`*** seems to be futile, hence ***`os`*** will mix variants of both OSes, phones. Let us continue the discussion in the thread [\"Freq counts from Train &amp; Test datasets\"][1], see his spreadsheet.\n\n(FYI, given this is from China in Nov 2017, just before Singles Day, the best prior estimate for smartphone distribution might be \"top phones in China 2017\" or else \"2016\". But anyway that seems useless if ***`os`*** combines device and Android fragmentation. Anyway let's continue all this in the other thread.)\n\n  [1]: https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/52201",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 301785,
      "author_name": "smcinerney",
      "author_url": "",
      "post_date": "03/23/2018 08:44:36",
      "content": "<p>I wondered the same, but unmasking <strong><em><code>device</code></em></strong> seems to be futile, hence <strong><em><code>os</code></em></strong> will mix variants of both OSes, phones. Let us continue the discussion in the thread <a href=\"https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/52201\">\"Freq counts from Train &amp; Test datasets\"</a>, see his spreadsheet.</p>\n\n<p>(FYI, given this is from China in Nov 2017, just before Singles Day, the best prior estimate for smartphone distribution might be \"top phones in China 2017\" or else \"2016\". But anyway that seems useless if <strong><em><code>os</code></em></strong> combines device and Android fragmentation. Anyway let's continue all this in the other thread.)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "295762": "Is there any map to use for decoding the phone codec in data?\nThank you!",
    "301785": "I wondered the same, but unmasking ***`device`*** seems to be futile, hence ***`os`*** will mix variants of both OSes, phones. Let us continue the discussion in the thread [\"Freq counts from Train &amp; Test datasets\"][1], see his spreadsheet.\n\n(FYI, given this is from China in Nov 2017, just before Singles Day, the best prior estimate for smartphone distribution might be \"top phones in China 2017\" or else \"2016\". But anyway that seems useless if ***`os`*** combines device and Android fragmentation. Anyway let's continue all this in the other thread.)\n\n  [1]: https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/52201"
  },
  "source": "meta"
}