{
  "id": 93041,
  "title": "Hmm..",
  "url": "/competitions/LANL-Earthquake-Prediction/discussion/93041",
  "author_name": "",
  "post_date": "2019-05-22T15:14:31.708337800Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p><a href=\"https://www.youtube.com/watch?v=mPaZmrpy7ng\">https://www.youtube.com/watch?v=mPaZmrpy7ng</a></p>",
  "messages": [
    {
      "id": "535249",
      "postDate": "05/22/2019 15:14:31",
      "content": "<p><a href=\"https://www.youtube.com/watch?v=mPaZmrpy7ng\">https://www.youtube.com/watch?v=mPaZmrpy7ng</a></p>",
      "rawMarkdown": "https://www.youtube.com/watch?v=mPaZmrpy7ng",
      "votes": null
    },
    {
      "id": "535259",
      "postDate": "05/22/2019 15:21:50",
      "content": "<p>If we can predict earthquakes with machine learning, it is a great contribution to humanity. So let's do our best for this earthquake competition of kaggle :)</p>",
      "rawMarkdown": "If we can predict earthquakes with machine learning, it is a great contribution to humanity. So let's do our best for this earthquake competition of kaggle :)",
      "votes": null
    },
    {
      "id": "535297",
      "postDate": "05/22/2019 16:28:17",
      "content": "<p>From the corresponding paper: \"Continuous chatter of the Cascadia subduction zone revealed by machine learning\"</p>\n\n<blockquote>\n  <p>Interestingly, in the case of the Cascadia slow earthquakes, the statistical features identified as most important by the machine-learning model are very similar to those found in the laboratory. The most important features are related to the power of the signal, but with extreme values removed, using interquantile ranges instead of higher-order moments. Extreme values are, in general, related to anthropogenic noise or small earthquakes, and quantiles help eliminate them and enable the model to focus on the acoustic power of the continuous background tremor.</p>\n</blockquote>",
      "rawMarkdown": "From the corresponding paper: \"Continuous chatter of the Cascadia subduction zone revealed by machine learning\"\n\n&gt; Interestingly, in the case of the Cascadia slow earthquakes, the statistical features identified as most important by the machine-learning model are very similar to those found in the laboratory. The most important features are related to the power of the signal, but with extreme values removed, using interquantile ranges instead of higher-order moments. Extreme values are, in general, related to anthropogenic noise or small earthquakes, and quantiles help eliminate them and enable the model to focus on the acoustic power of the continuous background tremor.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 535259,
      "author_name": "dhaqui",
      "author_url": "",
      "post_date": "05/22/2019 15:21:50",
      "content": "<p>If we can predict earthquakes with machine learning, it is a great contribution to humanity. So let's do our best for this earthquake competition of kaggle :)</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 535297,
      "author_name": "mcsnappy",
      "author_url": "",
      "post_date": "05/22/2019 16:28:17",
      "content": "<p>From the corresponding paper: \"Continuous chatter of the Cascadia subduction zone revealed by machine learning\"</p>\n\n<blockquote>\n  <p>Interestingly, in the case of the Cascadia slow earthquakes, the statistical features identified as most important by the machine-learning model are very similar to those found in the laboratory. The most important features are related to the power of the signal, but with extreme values removed, using interquantile ranges instead of higher-order moments. Extreme values are, in general, related to anthropogenic noise or small earthquakes, and quantiles help eliminate them and enable the model to focus on the acoustic power of the continuous background tremor.</p>\n</blockquote>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "535249": "https://www.youtube.com/watch?v=mPaZmrpy7ng",
    "535259": "If we can predict earthquakes with machine learning, it is a great contribution to humanity. So let's do our best for this earthquake competition of kaggle :)",
    "535297": "From the corresponding paper: \"Continuous chatter of the Cascadia subduction zone revealed by machine learning\"\n\n&gt; Interestingly, in the case of the Cascadia slow earthquakes, the statistical features identified as most important by the machine-learning model are very similar to those found in the laboratory. The most important features are related to the power of the signal, but with extreme values removed, using interquantile ranges instead of higher-order moments. Extreme values are, in general, related to anthropogenic noise or small earthquakes, and quantiles help eliminate them and enable the model to focus on the acoustic power of the continuous background tremor."
  },
  "source": "meta"
}