{
  "id": 248829,
  "title": "combining submission for _study and _image parts of data",
  "url": "/competitions/siim-covid19-detection/discussion/248829",
  "author_name": "Sameer Mahajan",
  "post_date": "2021-06-25T06:16:24.822000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>We made individual submissions for _study and _image part of  the data as  the models are different. e.g. in submission for _study part we made a default label for _image and vice versa.</p>\n<p>Then we combined the results in another submission. However the leaderboard score for the combined submission was lower than that of individual submission. What are we missing?</p>\n<p>Also any insight into how leaderboard score is calculated so that it can be tracked easily during our experimentation?</p>",
  "messages": [
    {
      "id": 1364679,
      "postDate": "2021-06-25T06:16:24.823Z",
      "content": "<p>We made individual submissions for _study and _image part of  the data as  the models are different. e.g. in submission for _study part we made a default label for _image and vice versa.</p>\n<p>Then we combined the results in another submission. However the leaderboard score for the combined submission was lower than that of individual submission. What are we missing?</p>\n<p>Also any insight into how leaderboard score is calculated so that it can be tracked easily during our experimentation?</p>",
      "rawMarkdown": "We made individual submissions for _study and _image part of  the data as  the models are different. e.g. in submission for _study part we made a default label for _image and vice versa.\n\nThen we combined the results in another submission. However the leaderboard score for the combined submission was lower than that of individual submission. What are we missing?\n\nAlso any insight into how leaderboard score is calculated so that it can be tracked easily during our experimentation?"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1364679": "We made individual submissions for _study and _image part of  the data as  the models are different. e.g. in submission for _study part we made a default label for _image and vice versa.\n\nThen we combined the results in another submission. However the leaderboard score for the combined submission was lower than that of individual submission. What are we missing?\n\nAlso any insight into how leaderboard score is calculated so that it can be tracked easily during our experimentation?"
  }
}