{
  "id": 10944,
  "title": "How was the train / test split created?",
  "url": "/competitions/seizure-prediction/discussion/10944",
  "author_name": "Mike Kim",
  "post_date": "2014-11-18T00:33:44.190000",
  "votes": 0,
  "comment_count": 0,
  "views": 559,
  "content": "<p>I have two questions regarding the dataset given the 0.05 to 0.10 AUC losses across the leaderboard.</p>\n<p>1) How was the split that created train and test data generated?</p>\n<p>2) How was the test data split into public and private leaderboard?</p>\n<p>This is important since I'll be doing some extra work involving this dataset and I'd like to correctly apply CV to match the leaderboard.</p>\n<p>Thanks</p>",
  "messages": [
    {
      "id": 58242,
      "postDate": "2014-11-18T00:33:44.190Z",
      "content": "<p>I have two questions regarding the dataset given the 0.05 to 0.10 AUC losses across the leaderboard.</p>\n<p>1) How was the split that created train and test data generated?</p>\n<p>2) How was the test data split into public and private leaderboard?</p>\n<p>This is important since I'll be doing some extra work involving this dataset and I'd like to correctly apply CV to match the leaderboard.</p>\n<p>Thanks</p>"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "58242": ""
  }
}