{
  "id": 5099,
  "title": "Data Description",
  "url": "/competitions/belkin-energy-disaggregation-competition/discussion/5099",
  "author_name": "",
  "post_date": "2013-07-15T03:44:08.923Z",
  "votes": null,
  "comment_count": 2,
  "views": 1784,
  "content": "<p>Hi,</p>\r\n<p>So I was looking through the data files and had a few questions. Sorry don't come from a EE background. :P</p>\r\n<p>Buffer.HF</p>\r\n<ol>\r\n<li>What do 4096 rows represent? </li><li>What are FFT vectors? </li></ol>\r\n<p>Thanks!</p>",
  "messages": [
    {
      "id": "27208",
      "postDate": "07/15/2013 03:44:08",
      "content": "<p>Hi,</p>\r\n<p>So I was looking through the data files and had a few questions. Sorry don't come from a EE background. :P</p>\r\n<p>Buffer.HF</p>\r\n<ol>\r\n<li>What do 4096 rows represent? </li><li>What are FFT vectors? </li></ol>\r\n<p>Thanks!</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "27210",
      "postDate": "07/15/2013 04:51:31",
      "content": "<p>Each row is a specific frequency component.&nbsp;</p>\r\n<p>FFT vector is a column of the power in each frequency component.&nbsp;<a>http://en.wikipedia.org/wiki/Fast_Fourier_transform</a></p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "27212",
      "postDate": "07/15/2013 05:33:42",
      "content": "<p>4096 is the number of frequency divisions. So in case of the data provided, across a spectrum of 0-1Mhz, there are 4096 divisions. In other words, each &quot;bin&quot; or index as you may call it, holds the amount of energy at that frequency. 1Mhz/4096 = 244.14Hz,\r\n thus, each bin from 0-4095 represents energy averaged over roughly 244.14Hz.</p>\r\n<p>If a CFL lamp produces a lot of &quot;noise&quot; or energy at 60kHz when operational, then you should see this at around 60000/244.14 = 245th bin or index in the vector. Since the noise is not very sharp and at an exact frequency, the energy generally spans many\r\n tens or hundreds of Hz or many bins.</p>\r\n<p>Each FFT vector is 4096 bins long and is a result of a short time Fourier transform of 8192 time domain sample points.</p>\r\n<p>Matlab and National Instruments have a great primer on FFT on their websites along with sample codes to quickly understand FFT. It is very commonly used across all science disciplines. I am too not of EE background :)</p>\r\n<p>Sidhant</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 27210,
      "author_name": "ibakasura",
      "author_url": "",
      "post_date": "07/15/2013 04:51:31",
      "content": "<p>Each row is a specific frequency component.&nbsp;</p>\r\n<p>FFT vector is a column of the power in each frequency component.&nbsp;<a>http://en.wikipedia.org/wiki/Fast_Fourier_transform</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 27212,
      "author_name": "sidhantgupta",
      "author_url": "",
      "post_date": "07/15/2013 05:33:42",
      "content": "<p>4096 is the number of frequency divisions. So in case of the data provided, across a spectrum of 0-1Mhz, there are 4096 divisions. In other words, each &quot;bin&quot; or index as you may call it, holds the amount of energy at that frequency. 1Mhz/4096 = 244.14Hz,\r\n thus, each bin from 0-4095 represents energy averaged over roughly 244.14Hz.</p>\r\n<p>If a CFL lamp produces a lot of &quot;noise&quot; or energy at 60kHz when operational, then you should see this at around 60000/244.14 = 245th bin or index in the vector. Since the noise is not very sharp and at an exact frequency, the energy generally spans many\r\n tens or hundreds of Hz or many bins.</p>\r\n<p>Each FFT vector is 4096 bins long and is a result of a short time Fourier transform of 8192 time domain sample points.</p>\r\n<p>Matlab and National Instruments have a great primer on FFT on their websites along with sample codes to quickly understand FFT. It is very commonly used across all science disciplines. I am too not of EE background :)</p>\r\n<p>Sidhant</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "27208": "",
    "27210": "",
    "27212": ""
  },
  "source": "meta"
}