{
  "id": 326252,
  "title": "Fun viz. of relationship between meter reading and anomaly",
  "url": "/competitions/energy-anomaly-detection/discussion/326252",
  "author_name": "Yoda",
  "post_date": "2022-05-21T04:22:07.219000",
  "votes": 6,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Meter reading appears the most significant of all features. For viz. purpose, a density plot (credits.- <a href=\"https://www.kaggle.com/headsortails\" target=\"_blank\">@headsortails</a> ) is what I feel provides a great intuition on why it's a significant driver. <br>\nThe spike at 0 is what's creating a strong separation. Almost 45% of all anomalies have a log(meter reading) = 0. Also, of all readings with log(meter_reading) = 0, ~100% are anomalous.</p>\n<p><img src=\"https://raw.githubusercontent.com/rohannanaware/Machine-Learning/master/Kaggle/LEAD/Screenshot%202022-05-21%20at%209.43.58%20AM.png\" alt=\"Markdown Monster icon\"></p>",
  "messages": [
    {
      "id": 1796723,
      "postDate": "2022-05-21T04:22:07.220Z",
      "content": "<p>Meter reading appears the most significant of all features. For viz. purpose, a density plot (credits.- <a href=\"https://www.kaggle.com/headsortails\" target=\"_blank\">@headsortails</a> ) is what I feel provides a great intuition on why it's a significant driver. <br>\nThe spike at 0 is what's creating a strong separation. Almost 45% of all anomalies have a log(meter reading) = 0. Also, of all readings with log(meter_reading) = 0, ~100% are anomalous.</p>\n<p><img src=\"https://raw.githubusercontent.com/rohannanaware/Machine-Learning/master/Kaggle/LEAD/Screenshot%202022-05-21%20at%209.43.58%20AM.png\" alt=\"Markdown Monster icon\"></p>",
      "rawMarkdown": "Meter reading appears the most significant of all features. For viz. purpose, a density plot (credits.- @headsortails ) is what I feel provides a great intuition on why it's a significant driver. \nThe spike at 0 is what's creating a strong separation. Almost 45% of all anomalies have a log(meter reading) = 0. Also, of all readings with log(meter_reading) = 0, ~100% are anomalous.\n\n<img src=\"https://raw.githubusercontent.com/rohannanaware/Machine-Learning/master/Kaggle/LEAD/Screenshot%202022-05-21%20at%209.43.58%20AM.png\"\n     alt=\"Markdown Monster icon\"\n     style=\"float: left; margin-right: 10px;\" />\n",
      "votes": 6
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1796723": "Meter reading appears the most significant of all features. For viz. purpose, a density plot (credits.- @headsortails ) is what I feel provides a great intuition on why it's a significant driver. \nThe spike at 0 is what's creating a strong separation. Almost 45% of all anomalies have a log(meter reading) = 0. Also, of all readings with log(meter_reading) = 0, ~100% are anomalous.\n\n<img src=\"https://raw.githubusercontent.com/rohannanaware/Machine-Learning/master/Kaggle/LEAD/Screenshot%202022-05-21%20at%209.43.58%20AM.png\"\n     alt=\"Markdown Monster icon\"\n     style=\"float: left; margin-right: 10px;\" />\n"
  }
}