{
  "id": 19743,
  "title": "Quantile Regression Forest",
  "url": "/competitions/second-annual-data-science-bowl/discussion/19743",
  "author_name": "Icedragon",
  "post_date": "2016-03-23T18:01:12.103000",
  "votes": 1,
  "comment_count": 0,
  "views": 401,
  "content": "<p>Hi,</p>\n\n<p>I got ~0.033 without getting more data from randomly selected images and getting basic statistics out of them like the information at the beginning and darkest or lightest pixel by fitting a quantile regression forest with R-package quantregForest. I made a grid with 10000 quantiles and transformed this quantiles to a cdf.</p>\n\n<p>Probably more would have been possible with more sophisticated models for getting more informations and better features out of the images. \nAnybody else used this? \nI unfortunately did not fit the model to the test data so I am not in the leaderboard. </p>",
  "messages": [
    {
      "id": 112776,
      "postDate": "2016-03-23T18:01:12.103Z",
      "content": "<p>Hi,</p>\n\n<p>I got ~0.033 without getting more data from randomly selected images and getting basic statistics out of them like the information at the beginning and darkest or lightest pixel by fitting a quantile regression forest with R-package quantregForest. I made a grid with 10000 quantiles and transformed this quantiles to a cdf.</p>\n\n<p>Probably more would have been possible with more sophisticated models for getting more informations and better features out of the images. \nAnybody else used this? \nI unfortunately did not fit the model to the test data so I am not in the leaderboard. </p>",
      "rawMarkdown": "Hi,\r\n\r\nI got ~0.033 without getting more data from randomly selected images and getting basic statistics out of them like the information at the beginning and darkest or lightest pixel by fitting a quantile regression forest with R-package quantregForest. I made a grid with 10000 quantiles and transformed this quantiles to a cdf.\r\n\r\nProbably more would have been possible with more sophisticated models for getting more informations and better features out of the images. \r\nAnybody else used this? \r\nI unfortunately did not fit the model to the test data so I am not in the leaderboard. ",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "112776": "Hi,\r\n\r\nI got ~0.033 without getting more data from randomly selected images and getting basic statistics out of them like the information at the beginning and darkest or lightest pixel by fitting a quantile regression forest with R-package quantregForest. I made a grid with 10000 quantiles and transformed this quantiles to a cdf.\r\n\r\nProbably more would have been possible with more sophisticated models for getting more informations and better features out of the images. \r\nAnybody else used this? \r\nI unfortunately did not fit the model to the test data so I am not in the leaderboard. "
  }
}