{
  "id": 34864,
  "title": "Does the score correct for random guessing?",
  "url": "/competitions/digit-recognizer/discussion/34864",
  "author_name": "Rex",
  "post_date": "2017-06-17T03:11:11.918000",
  "votes": 0,
  "comment_count": 1,
  "views": null,
  "content": "<p>I made a test submission using prediction from a very naive linear logistic regression. I didn't really expect to do well, but I'm still surprised that I had a score that is dead 0 :D. If the score was not corrected by random guessing, I guess I actually did better than guessing? lol</p>\n\n<p>The model has about 85% accuracy on cross-validated samples, so yeah, I'm a bit surprised. Anyone else has the same issue?</p>",
  "messages": [
    {
      "id": 193606,
      "postDate": "2017-06-17T03:47:48.140Z",
      "content": "<p>OK. I have figured it out. The \"Label\" column has to be explicitly coerced to integer. </p>",
      "rawMarkdown": "OK. I have figured it out. The \"Label\" column has to be explicitly coerced to integer. "
    },
    {
      "id": 193602,
      "postDate": "2017-06-17T03:11:11.920Z",
      "content": "<p>I made a test submission using prediction from a very naive linear logistic regression. I didn't really expect to do well, but I'm still surprised that I had a score that is dead 0 :D. If the score was not corrected by random guessing, I guess I actually did better than guessing? lol</p>\n\n<p>The model has about 85% accuracy on cross-validated samples, so yeah, I'm a bit surprised. Anyone else has the same issue?</p>",
      "rawMarkdown": "I made a test submission using prediction from a very naive linear logistic regression. I didn't really expect to do well, but I'm still surprised that I had a score that is dead 0 :D. If the score was not corrected by random guessing, I guess I actually did better than guessing? lol\n\nThe model has about 85% accuracy on cross-validated samples, so yeah, I'm a bit surprised. Anyone else has the same issue?"
    }
  ],
  "comments": [
    {
      "id": 193606,
      "author_name": "Rex",
      "author_url": "",
      "post_date": "2017-06-17T03:47:48.140000",
      "content": "<p>OK. I have figured it out. The \"Label\" column has to be explicitly coerced to integer. </p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "193606": "OK. I have figured it out. The \"Label\" column has to be explicitly coerced to integer. ",
    "193602": "I made a test submission using prediction from a very naive linear logistic regression. I didn't really expect to do well, but I'm still surprised that I had a score that is dead 0 :D. If the score was not corrected by random guessing, I guess I actually did better than guessing? lol\n\nThe model has about 85% accuracy on cross-validated samples, so yeah, I'm a bit surprised. Anyone else has the same issue?"
  }
}