{ "id": 107118, "title": "Robust ensemble methods", "url": "/competitions/aptos2019-blindness-detection/discussion/107118", "author_name": "", "post_date": "2019-09-02T11:59:00.756961900Z", "votes": 11, "comment_count": 2, "views": 0, "content": "

Simple average is the way that is vulnerable to outliers. There are several robust ensemble methods for outliers.\nex)\nα-Trimmed mean\nMedian

\n\n

Modern Methods for Robust Regression

\n\n

[Additional Ensemble method I created for this comp]\nCalculate the average of the histogram and determine the round threshold based on the result.

", "messages": [ { "id": "615816", "postDate": "09/02/2019 11:59:00", "content": "

Simple average is the way that is vulnerable to outliers. There are several robust ensemble methods for outliers.\nex)\nα-Trimmed mean\nMedian

\n\n

Modern Methods for Robust Regression

\n\n

[Additional Ensemble method I created for this comp]\nCalculate the average of the histogram and determine the round threshold based on the result.

", "rawMarkdown": "Simple average is the way that is vulnerable to outliers. There are several robust ensemble methods for outliers.\nex)\nα-Trimmed mean\nMedian\n\n[Modern Methods for Robust Regression](https://books.google.co.jp/books?id=ce5yKCu8HRoC&pg=PA12&lpg=PA12&dq=modern+regression+methods+trim+mean&source=bl&ots=-ZTVmdO6Co&sig=ACfU3U3GV1U-mITQoDlr1j66DofSIlnE6g&hl=ja&sa=X&ved=2ahUKEwiDgNS9ibLkAhXnFqYKHaQBDuMQ6AEwAHoECAcQAQ#v=onepage&q=modern%20regression%20methods%20trim%20mean&f=false)\n\n[Additional Ensemble method I created for this comp]\nCalculate the average of the histogram and determine the round threshold based on the result.", "votes": null }, { "id": "615959", "postDate": "09/02/2019 14:44:33", "content": "

Thank you. Hope I can still make it

", "rawMarkdown": "Thank you. Hope I can still make it", "votes": null }, { "id": "618800", "postDate": "09/05/2019 14:02:06", "content": "

tri-mean seems to be quite effective for me - I'd imagine it would also work well with larger ensembles if people are not looking to use weighting schemes.

", "rawMarkdown": "tri-mean seems to be quite effective for me - I'd imagine it would also work well with larger ensembles if people are not looking to use weighting schemes.", "votes": null } ], "comments": [ { "id": 615959, "author_name": "lanhhv84", "author_url": "", "post_date": "09/02/2019 14:44:33", "content": "

Thank you. Hope I can still make it

", "votes": null, "replies": [] }, { "id": 618800, "author_name": "", "author_url": "", "post_date": "09/05/2019 14:02:06", "content": "

tri-mean seems to be quite effective for me - I'd imagine it would also work well with larger ensembles if people are not looking to use weighting schemes.

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "615816": "Simple average is the way that is vulnerable to outliers. There are several robust ensemble methods for outliers.\nex)\nα-Trimmed mean\nMedian\n\n[Modern Methods for Robust Regression](https://books.google.co.jp/books?id=ce5yKCu8HRoC&pg=PA12&lpg=PA12&dq=modern+regression+methods+trim+mean&source=bl&ots=-ZTVmdO6Co&sig=ACfU3U3GV1U-mITQoDlr1j66DofSIlnE6g&hl=ja&sa=X&ved=2ahUKEwiDgNS9ibLkAhXnFqYKHaQBDuMQ6AEwAHoECAcQAQ#v=onepage&q=modern%20regression%20methods%20trim%20mean&f=false)\n\n[Additional Ensemble method I created for this comp]\nCalculate the average of the histogram and determine the round threshold based on the result.", "615959": "Thank you. Hope I can still make it", "618800": "tri-mean seems to be quite effective for me - I'd imagine it would also work well with larger ensembles if people are not looking to use weighting schemes." }, "source": "meta" }