{
  "id": 18684,
  "title": "Predictions - spread",
  "url": "/competitions/second-annual-data-science-bowl/discussion/18684",
  "author_name": "WD",
  "post_date": "2016-02-01T16:29:36.763000",
  "votes": 0,
  "comment_count": 0,
  "views": 334,
  "content": "<p>I am running a linear regression setup (RSME metric), and was looking at the predictions on the training set. I spotted a pattern that nearly all predictions for systole and diastole are smaller than the actual values. Is that possible - or is that a bug somewher in my model? Even with using a RSME metric - I would have thought that the errors in practice would be both positive and negative, and that there would be roughly an equal number of predictions that would be too high and too low.... let me know any thoughts </p>",
  "messages": [
    {
      "id": 106515,
      "postDate": "2016-02-01T16:29:36.763Z",
      "content": "<p>I am running a linear regression setup (RSME metric), and was looking at the predictions on the training set. I spotted a pattern that nearly all predictions for systole and diastole are smaller than the actual values. Is that possible - or is that a bug somewher in my model? Even with using a RSME metric - I would have thought that the errors in practice would be both positive and negative, and that there would be roughly an equal number of predictions that would be too high and too low.... let me know any thoughts </p>",
      "rawMarkdown": "I am running a linear regression setup (RSME metric), and was looking at the predictions on the training set. I spotted a pattern that nearly all predictions for systole and diastole are smaller than the actual values. Is that possible - or is that a bug somewher in my model? Even with using a RSME metric - I would have thought that the errors in practice would be both positive and negative, and that there would be roughly an equal number of predictions that would be too high and too low.... let me know any thoughts "
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "106515": "I am running a linear regression setup (RSME metric), and was looking at the predictions on the training set. I spotted a pattern that nearly all predictions for systole and diastole are smaller than the actual values. Is that possible - or is that a bug somewher in my model? Even with using a RSME metric - I would have thought that the errors in practice would be both positive and negative, and that there would be roughly an equal number of predictions that would be too high and too low.... let me know any thoughts "
  }
}