{
  "id": 86458,
  "title": "Handling the gap in recordings",
  "url": "/competitions/LANL-Earthquake-Prediction/discussion/86458",
  "author_name": "Mahir Jain",
  "post_date": "2019-03-24T06:26:51.257000",
  "votes": 3,
  "comment_count": 0,
  "views": 0,
  "content": "<p>The information provided states that the data has been recorded in bins of 4096 elements, with a 12 microsecond gap in between bins.</p>\n\n<p>Is it possible that not accounting for this, and just splitting training data into segments of 150,000 points each is causing a problem? What could be the effects of accounting for this gap and would it be better than the simple splitting approach? </p>\n\n<p>Looking for a clear, logical explanation on this...</p>",
  "messages": [
    {
      "id": 498983,
      "postDate": "2019-03-24T06:26:51.257Z",
      "content": "<p>The information provided states that the data has been recorded in bins of 4096 elements, with a 12 microsecond gap in between bins.</p>\n\n<p>Is it possible that not accounting for this, and just splitting training data into segments of 150,000 points each is causing a problem? What could be the effects of accounting for this gap and would it be better than the simple splitting approach? </p>\n\n<p>Looking for a clear, logical explanation on this...</p>",
      "rawMarkdown": "The information provided states that the data has been recorded in bins of 4096 elements, with a 12 microsecond gap in between bins.\n\nIs it possible that not accounting for this, and just splitting training data into segments of 150,000 points each is causing a problem? What could be the effects of accounting for this gap and would it be better than the simple splitting approach? \n\nLooking for a clear, logical explanation on this...",
      "votes": 3
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "498983": "The information provided states that the data has been recorded in bins of 4096 elements, with a 12 microsecond gap in between bins.\n\nIs it possible that not accounting for this, and just splitting training data into segments of 150,000 points each is causing a problem? What could be the effects of accounting for this gap and would it be better than the simple splitting approach? \n\nLooking for a clear, logical explanation on this..."
  }
}