{
  "id": 5125,
  "title": "Finding feature vectors for timestamps in sample submission",
  "url": "/competitions/belkin-energy-disaggregation-competition/discussion/5125",
  "author_name": "",
  "post_date": "2013-07-17T05:21:54.950Z",
  "votes": null,
  "comment_count": 1,
  "views": 1117,
  "content": "<p>Hi,</p>\r\n<p>I did some simple tests and it looks like the timestamps given in the sample submission file, for H1, does not match with any of the timestamps in any of the testing files for H1. I assume this is done on purpose. Appreciate if anyone can give me some pointers\r\n on how to go about creating feature vectors for the submission timestamps given they dont match with anything. Am I missing something? Are others seeing the same thing?</p>\r\n<p>Thanks</p>\r\n<p>Rahul</p>",
  "messages": [
    {
      "id": "27326",
      "postDate": "07/17/2013 05:21:54",
      "content": "<p>Hi,</p>\r\n<p>I did some simple tests and it looks like the timestamps given in the sample submission file, for H1, does not match with any of the timestamps in any of the testing files for H1. I assume this is done on purpose. Appreciate if anyone can give me some pointers\r\n on how to go about creating feature vectors for the submission timestamps given they dont match with anything. Am I missing something? Are others seeing the same thing?</p>\r\n<p>Thanks</p>\r\n<p>Rahul</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "27329",
      "postDate": "07/17/2013 08:02:29",
      "content": "<p>I guess you mean the timestamps in Buffer.TimeTicksHF. Those timeticks refered to the instants at which the FFT coefficients are computed, and those coefficients are computed over a time window. That's the reason why those timestamps are not the same as\r\n in the submission file.</p>\r\n<p>I suggest you to just consider that an event at a given instant t will have an influence on the next FFT vector computed immediately after t.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 27329,
      "author_name": "joseleiva",
      "author_url": "",
      "post_date": "07/17/2013 08:02:29",
      "content": "<p>I guess you mean the timestamps in Buffer.TimeTicksHF. Those timeticks refered to the instants at which the FFT coefficients are computed, and those coefficients are computed over a time window. That's the reason why those timestamps are not the same as\r\n in the submission file.</p>\r\n<p>I suggest you to just consider that an event at a given instant t will have an influence on the next FFT vector computed immediately after t.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "27326": "",
    "27329": ""
  },
  "source": "meta"
}