{ "id": 278815, "title": "[WARNING]Don't select a public notebook that doesn't infer private data as the final submission", "url": "/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/278815", "author_name": "", "post_date": "2021-10-15T18:20:32.887471100Z", "votes": 9, "comment_count": 11, "views": 0, "content": "

[WARNING] Don't select a public notebook that doesn't infer private data as the final submission.
\nIf you select it, you will get a very low score and be a very low rank in the private leaderboard
\nEven if the score looks low, please select a public notebook that infers private data. You might get lucky and have higher private ranking in the end.

", "messages": [ { "id": "1545986", "postDate": "10/15/2021 18:20:32", "content": "

[WARNING] Don't select a public notebook that doesn't infer private data as the final submission.
\nIf you select it, you will get a very low score and be a very low rank in the private leaderboard
\nEven if the score looks low, please select a public notebook that infers private data. You might get lucky and have higher private ranking in the end.

", "rawMarkdown": "**[WARNING] Don't select a public notebook that doesn't infer private data as the final submission.**\nIf you select it, you will get a very low score and be a very low rank in the private leaderboard\nEven if the score looks low, please select a public notebook that infers private data. You might get lucky and have higher private ranking in the end.", "votes": null }, { "id": "1546221", "postDate": "10/15/2021 23:05:03", "content": "

This is good advice. I'm not participating in this competition but i see that many teams have a public score of 0.732. If those teams forked the common high scoring public notebooks, all those teams will get private score of 0.000!

", "rawMarkdown": "This is good advice. I'm not participating in this competition but i see that many teams have a public score of 0.732. If those teams forked the common high scoring public notebooks, all those teams will get private score of 0.000!", "votes": null }, { "id": "1546230", "postDate": "10/15/2021 23:18:52", "content": "
\n

You might get lucky

\n
\n

The entire sentiment of the 1500 participants right now

", "rawMarkdown": "> You might get lucky\n\nThe entire sentiment of the 1500 participants right now", "votes": null }, { "id": "1546232", "postDate": "10/15/2021 23:21:28", "content": "

True but if you select a public notebook that doesn't make predictions for private LB, then you won't get lucky. You'll get private score 0.000

", "rawMarkdown": "True but if you select a public notebook that doesn't make predictions for private LB, then you won't get lucky. You'll get private score 0.000", "votes": null }, { "id": "1546247", "postDate": "10/15/2021 23:59:02", "content": "

I read this as \"I might not wind up in the bottom 10% after all.\" Fingers crossed! 😂😭

", "rawMarkdown": "I read this as \"I might not wind up in the bottom 10% after all.\" Fingers crossed! 😂😭", "votes": null }, { "id": "1546252", "postDate": "10/16/2021 00:02:45", "content": "

haha, you will probably shake up a lot. I think many teams will have private score 0.000

", "rawMarkdown": "haha, you will probably shake up a lot. I think many teams will have private score 0.000", "votes": null }, { "id": "1546312", "postDate": "10/16/2021 02:37:23", "content": "

it would have been great if you had participated @cdeotte, if you see the discussions you can see how difficult it was for our models to learn in this competition.
\nYour experience may have helped us. But lets see what happens in the private leaderboard,I dont think the winner may get as high scores as public results.

", "rawMarkdown": "it would have been great if you had participated @cdeotte, if you see the discussions you can see how difficult it was for our models to learn in this competition.\nYour experience may have helped us. But lets see what happens in the private leaderboard,I dont think the winner may get as high scores as public results.", "votes": null }, { "id": "1546480", "postDate": "10/16/2021 07:43:17", "content": "

good !!!!!!!

", "rawMarkdown": "good !!!!!!!", "votes": null }, { "id": "1546548", "postDate": "10/16/2021 09:26:57", "content": "

Actually the score 0.0 for AUC would mean perfect prediction of the opposite classes (0 instead of 1, and 1 instead of 0). I guess we should rather expect something around 0.5, right? 😀

", "rawMarkdown": "Actually the score 0.0 for AUC would mean perfect prediction of the opposite classes (0 instead of 1, and 1 instead of 0). I guess we should rather expect something around 0.5, right? 😀", "votes": null }, { "id": "1546685", "postDate": "10/16/2021 12:32:24", "content": "

Yes, you are correct. Good point.

", "rawMarkdown": "Yes, you are correct. Good point.", "votes": null }, { "id": "1549279", "postDate": "10/18/2021 22:08:33", "content": "

@daverogers1 Well done Dave! You got top 44%!

", "rawMarkdown": "daverogers1 Well done Dave! You got top 44%!", "votes": null }, { "id": "1549280", "postDate": "10/18/2021 22:09:34", "content": "

Great job @authman and @gunesevitan ! Congrats on Gold finish!

", "rawMarkdown": "Great job @authman and @gunesevitan ! Congrats on Gold finish!", "votes": null } ], "comments": [ { "id": 1546221, "author_name": "cdeotte", "author_url": "", "post_date": "10/15/2021 23:05:03", "content": "

This is good advice. I'm not participating in this competition but i see that many teams have a public score of 0.732. If those teams forked the common high scoring public notebooks, all those teams will get private score of 0.000!

", "votes": null, "replies": [ { "id": 1546247, "author_name": "daverogers1", "author_url": "", "post_date": "10/15/2021 23:59:02", "content": "

I read this as \"I might not wind up in the bottom 10% after all.\" Fingers crossed! 😂😭

", "votes": null, "replies": [] }, { "id": 1546252, "author_name": "cdeotte", "author_url": "", "post_date": "10/16/2021 00:02:45", "content": "

haha, you will probably shake up a lot. I think many teams will have private score 0.000

", "votes": null, "replies": [] }, { "id": 1546548, "author_name": "mikecho", "author_url": "", "post_date": "10/16/2021 09:26:57", "content": "

Actually the score 0.0 for AUC would mean perfect prediction of the opposite classes (0 instead of 1, and 1 instead of 0). I guess we should rather expect something around 0.5, right? 😀

", "votes": null, "replies": [] }, { "id": 1546685, "author_name": "cdeotte", "author_url": "", "post_date": "10/16/2021 12:32:24", "content": "

Yes, you are correct. Good point.

", "votes": null, "replies": [] }, { "id": 1549279, "author_name": "cdeotte", "author_url": "", "post_date": "10/18/2021 22:08:33", "content": "

@daverogers1 Well done Dave! You got top 44%!

", "votes": null, "replies": [] } ] }, { "id": 1546230, "author_name": "authman", "author_url": "", "post_date": "10/15/2021 23:18:52", "content": "
\n

You might get lucky

\n
\n

The entire sentiment of the 1500 participants right now

", "votes": null, "replies": [ { "id": 1546232, "author_name": "cdeotte", "author_url": "", "post_date": "10/15/2021 23:21:28", "content": "

True but if you select a public notebook that doesn't make predictions for private LB, then you won't get lucky. You'll get private score 0.000

", "votes": null, "replies": [] }, { "id": 1546312, "author_name": "mrinath", "author_url": "", "post_date": "10/16/2021 02:37:23", "content": "

it would have been great if you had participated @cdeotte, if you see the discussions you can see how difficult it was for our models to learn in this competition.
\nYour experience may have helped us. But lets see what happens in the private leaderboard,I dont think the winner may get as high scores as public results.

", "votes": null, "replies": [] }, { "id": 1549280, "author_name": "cdeotte", "author_url": "", "post_date": "10/18/2021 22:09:34", "content": "

Great job @authman and @gunesevitan ! Congrats on Gold finish!

", "votes": null, "replies": [] } ] }, { "id": 1546480, "author_name": "bowencui18", "author_url": "", "post_date": "10/16/2021 07:43:17", "content": "

good !!!!!!!

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1545986": "**[WARNING] Don't select a public notebook that doesn't infer private data as the final submission.**\nIf you select it, you will get a very low score and be a very low rank in the private leaderboard\nEven if the score looks low, please select a public notebook that infers private data. You might get lucky and have higher private ranking in the end.", "1546221": "This is good advice. I'm not participating in this competition but i see that many teams have a public score of 0.732. If those teams forked the common high scoring public notebooks, all those teams will get private score of 0.000!", "1546230": "> You might get lucky\n\nThe entire sentiment of the 1500 participants right now", "1546232": "True but if you select a public notebook that doesn't make predictions for private LB, then you won't get lucky. You'll get private score 0.000", "1546247": "I read this as \"I might not wind up in the bottom 10% after all.\" Fingers crossed! 😂😭", "1546252": "haha, you will probably shake up a lot. I think many teams will have private score 0.000", "1546312": "it would have been great if you had participated @cdeotte, if you see the discussions you can see how difficult it was for our models to learn in this competition.\nYour experience may have helped us. But lets see what happens in the private leaderboard,I dont think the winner may get as high scores as public results.", "1546480": "good !!!!!!!", "1546548": "Actually the score 0.0 for AUC would mean perfect prediction of the opposite classes (0 instead of 1, and 1 instead of 0). I guess we should rather expect something around 0.5, right? 😀", "1546685": "Yes, you are correct. Good point.", "1549279": "daverogers1 Well done Dave! You got top 44%!", "1549280": "Great job @authman and @gunesevitan ! Congrats on Gold finish!" }, "source": "meta" }