{
  "id": 5298,
  "title": "Gaussian Process model",
  "url": "/competitions/ams-2014-solar-energy-prediction-contest/discussion/5298",
  "author_name": "tinypig",
  "post_date": "2013-08-05T21:20:38.960000",
  "votes": 0,
  "comment_count": 0,
  "views": 969,
  "content": "<p>I'm interested in the use of kriging and related gaussian process regression techniques and think that they will be useful here (to interpolate the random fields at the new locations at the very least).</p>\n<p>Has anyone tried such approaches with much success? At first brush, the number of data points across time and space (16*9*11*5*5113) seems prohibitively large to calculate covariances with.</p>\n<p>any tips/tricks to limit covariance matrix size?</p>",
  "messages": [
    {
      "id": 28223,
      "postDate": "2013-08-05T21:20:38.960Z",
      "content": "<p>I'm interested in the use of kriging and related gaussian process regression techniques and think that they will be useful here (to interpolate the random fields at the new locations at the very least).</p>\n<p>Has anyone tried such approaches with much success? At first brush, the number of data points across time and space (16*9*11*5*5113) seems prohibitively large to calculate covariances with.</p>\n<p>any tips/tricks to limit covariance matrix size?</p>"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "28223": ""
  }
}