{ "id": 566699, "title": "Aspect ratio", "url": "/competitions/byu-locating-bacterial-flagellar-motors-2025/discussion/566699", "author_name": "", "post_date": "2025-03-06T15:35:11.221504100Z", "votes": 35, "comment_count": 7, "views": 0, "content": "

I noticed that tomography with an aspect ratio exceeding a certain value does not contain any motors at all.

\n

EDA notebook: https://www.kaggle.com/code/welshonionman/eda-aspect-ratio

", "messages": [ { "id": "3142672", "postDate": "03/06/2025 15:35:11", "content": "

I noticed that tomography with an aspect ratio exceeding a certain value does not contain any motors at all.

\n

EDA notebook: https://www.kaggle.com/code/welshonionman/eda-aspect-ratio

", "rawMarkdown": "I noticed that tomography with an aspect ratio exceeding a certain value does not contain any motors at all.\n\nEDA notebook: https://www.kaggle.com/code/welshonionman/eda-aspect-ratio", "votes": null }, { "id": "3143635", "postDate": "03/07/2025 12:27:34", "content": "

Good find, this could be a leak.

", "rawMarkdown": "Good find, this could be a leak.", "votes": null }, { "id": "3144874", "postDate": "03/09/2025 02:16:23", "content": "

@welshonionman, good catch, I will just add to the discussion (no intention to steal your thunder): \"\"

\n

table\"\"

", "rawMarkdown": "welshonionman, good catch, I will just add to the discussion (no intention to steal your thunder): ![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6259210%2F9e6555119eb97243930fee3df1c536bd%2Faspect%20ratio%20vs%20target.png?generation=1741486455286494&alt=media)\n\n**table**![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6259210%2Faf3738b4cb46b619c2e81c31ea59d55f%2FScreenshot%20from%202025-03-08%2021-14-33.png?generation=1741486499749495&alt=media)", "votes": null }, { "id": "3160552", "postDate": "03/26/2025 22:33:28", "content": "

i think this could be a leak, too.

", "rawMarkdown": "i think this could be a leak, too.", "votes": null }, { "id": "3160899", "postDate": "03/27/2025 09:10:52", "content": "

All of the 6.5um voxel spacing tomograms have no motors. I'm not sure how to use this.

", "rawMarkdown": "All of the 6.5um voxel spacing tomograms have no motors. I'm not sure how to use this.", "votes": null }, { "id": "3161208", "postDate": "03/27/2025 16:45:32", "content": "

We can confidently assure you that these are not leaks and while observing things like this is useful, it ultimately won't yield any benefit on the leaderboard.

", "rawMarkdown": "We can confidently assure you that these are not leaks and while observing things like this is useful, it ultimately won't yield any benefit on the leaderboard.", "votes": null }, { "id": "3161301", "postDate": "03/27/2025 18:06:43", "content": "

But I assume there are 6.5um voxel space tomograms with motors in the test set.

", "rawMarkdown": "But I assume there are 6.5um voxel space tomograms with motors in the test set.", "votes": null }, { "id": "3161972", "postDate": "03/28/2025 16:54:59", "content": "

You could assume that. We're not going to provide any additional information beyond what has already been shared about the contents of the test set.

\n

The point is that we assured the test set's integrity would not be compromised by elements of the metadata in the train set by not running an arbitrary train-test split. Expecting tomograms with 6.5um voxel space to contain similar features to the ones analyzed in this thread is reliant upon the assumption that the train set and test set come from the same population.

", "rawMarkdown": "You could assume that. We're not going to provide any additional information beyond what has already been shared about the contents of the test set. \n\nThe point is that we assured the test set's integrity would not be compromised by elements of the metadata in the train set by not running an arbitrary train-test split. Expecting tomograms with 6.5um voxel space to contain similar features to the ones analyzed in this thread is reliant upon the assumption that the train set and test set come from the same population.", "votes": null } ], "comments": [ { "id": 3143635, "author_name": "gunesevitan", "author_url": "", "post_date": "03/07/2025 12:27:34", "content": "

Good find, this could be a leak.

", "votes": null, "replies": [] }, { "id": 3144874, "author_name": "sergiosaharovskiy", "author_url": "", "post_date": "03/09/2025 02:16:23", "content": "

@welshonionman, good catch, I will just add to the discussion (no intention to steal your thunder): \"\"

\n

table\"\"

", "votes": null, "replies": [] }, { "id": 3160552, "author_name": "", "author_url": "", "post_date": "03/26/2025 22:33:28", "content": "

i think this could be a leak, too.

", "votes": null, "replies": [] }, { "id": 3160899, "author_name": "sakvaua", "author_url": "", "post_date": "03/27/2025 09:10:52", "content": "

All of the 6.5um voxel spacing tomograms have no motors. I'm not sure how to use this.

", "votes": null, "replies": [] }, { "id": 3161208, "author_name": "andrewjdarley", "author_url": "", "post_date": "03/27/2025 16:45:32", "content": "

We can confidently assure you that these are not leaks and while observing things like this is useful, it ultimately won't yield any benefit on the leaderboard.

", "votes": null, "replies": [ { "id": 3161301, "author_name": "sakvaua", "author_url": "", "post_date": "03/27/2025 18:06:43", "content": "

But I assume there are 6.5um voxel space tomograms with motors in the test set.

", "votes": null, "replies": [ { "id": 3161972, "author_name": "andrewjdarley", "author_url": "", "post_date": "03/28/2025 16:54:59", "content": "

You could assume that. We're not going to provide any additional information beyond what has already been shared about the contents of the test set.

\n

The point is that we assured the test set's integrity would not be compromised by elements of the metadata in the train set by not running an arbitrary train-test split. Expecting tomograms with 6.5um voxel space to contain similar features to the ones analyzed in this thread is reliant upon the assumption that the train set and test set come from the same population.

", "votes": null, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "3142672": "I noticed that tomography with an aspect ratio exceeding a certain value does not contain any motors at all.\n\nEDA notebook: https://www.kaggle.com/code/welshonionman/eda-aspect-ratio", "3143635": "Good find, this could be a leak.", "3144874": "welshonionman, good catch, I will just add to the discussion (no intention to steal your thunder): ![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6259210%2F9e6555119eb97243930fee3df1c536bd%2Faspect%20ratio%20vs%20target.png?generation=1741486455286494&alt=media)\n\n**table**![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6259210%2Faf3738b4cb46b619c2e81c31ea59d55f%2FScreenshot%20from%202025-03-08%2021-14-33.png?generation=1741486499749495&alt=media)", "3160552": "i think this could be a leak, too.", "3160899": "All of the 6.5um voxel spacing tomograms have no motors. I'm not sure how to use this.", "3161208": "We can confidently assure you that these are not leaks and while observing things like this is useful, it ultimately won't yield any benefit on the leaderboard.", "3161301": "But I assume there are 6.5um voxel space tomograms with motors in the test set.", "3161972": "You could assume that. We're not going to provide any additional information beyond what has already been shared about the contents of the test set. \n\nThe point is that we assured the test set's integrity would not be compromised by elements of the metadata in the train set by not running an arbitrary train-test split. Expecting tomograms with 6.5um voxel space to contain similar features to the ones analyzed in this thread is reliant upon the assumption that the train set and test set come from the same population." }, "source": "meta" }