{
  "id": 10465,
  "title": "Preprocessing",
  "url": "/competitions/seizure-prediction/discussion/10465",
  "author_name": "",
  "post_date": "2014-09-27T14:47:54.550Z",
  "votes": 3,
  "comment_count": 1,
  "views": 1821,
  "content": "<p>What preprocessing methods are effective for all targets? After reading discussions in previous seizure contest, http://www.kaggle.com/c/seizure-detection/forums/t/10079/features-for-seizure-detection and http://www.kaggle.com/c/seizure-detection/forums/t/10111/required-model-documentation-and-code, which are very useful (Many thanks!), I still cannot find a universally effective features/preprocessing methods, especially for Patient 2 (lower than 0.6 always). So far time correlation seems to be the best. What features do you find useful and stable? Thank you!</p>",
  "messages": [
    {
      "id": "55336",
      "postDate": "09/27/2014 14:47:54",
      "content": "<p>What preprocessing methods are effective for all targets? After reading discussions in previous seizure contest, http://www.kaggle.com/c/seizure-detection/forums/t/10079/features-for-seizure-detection and http://www.kaggle.com/c/seizure-detection/forums/t/10111/required-model-documentation-and-code, which are very useful (Many thanks!), I still cannot find a universally effective features/preprocessing methods, especially for Patient 2 (lower than 0.6 always). So far time correlation seems to be the best. What features do you find useful and stable? Thank you!</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "56219",
      "postDate": "10/18/2014 10:36:06",
      "content": "",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 56219,
      "author_name": "anirudha136",
      "author_url": "",
      "post_date": "10/18/2014 10:36:06",
      "content": "",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "55336": "",
    "56219": ""
  },
  "source": "meta"
}