![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4373580%2F9b753e10b8cfbdbe5b03f4b5549ac058%2FCaptura%20de%20pantalla%202026-01-29%20152420.png?generation=1769721911582925&alt=media\")
I just realized the IDs and names of the training data are sorted, my hypothesis was: 'Maybe it is possible to know which jaguar is which based on label proximity.' To my huge disdain, I was correct:
\n
To the competition hosts, please consider shuffling the test fold labels or change the numbering strategy to avoid this being an easy way to the top 10.
\n\nif you deem it adequate I can remove this post and the code proof but is important to acknowledge this kind of data problems to keep the competition fair.
", "messages": [ { "id": "3398888", "postDate": "01/29/2026 21:31:34", "content": "I just realized the IDs and names of the training data are sorted, my hypothesis was: 'Maybe it is possible to know which jaguar is which based on label proximity.' To my huge disdain, I was correct:
\n
To the competition hosts, please consider shuffling the test fold labels or change the numbering strategy to avoid this being an easy way to the top 10.
\n\nif you deem it adequate I can remove this post and the code proof but is important to acknowledge this kind of data problems to keep the competition fair.
", "rawMarkdown": "I just realized the IDs and names of the training data are sorted, my hypothesis was: 'Maybe it is possible to know which jaguar is which based on label proximity.' To my huge disdain, I was correct:\n\n\n\nTo the competition hosts, please consider shuffling the test fold labels or change the numbering strategy to avoid this being an easy way to the top 10.\n\n[here is a link to my tests.](https://www.kaggle.com/code/elcientfico/ordering-leakage-test)\n\nif you deem it adequate I can remove this post and the code proof but is important to acknowledge this kind of data problems to keep the competition fair.", "votes": null }, { "id": "3398904", "postDate": "01/29/2026 21:58:19", "content": "Thanks for the feedback! Please don't feel obliged to delete the post. This is a research and education competition and I am iterating on its setup.
", "rawMarkdown": "Thanks for the feedback! Please don't feel obliged to delete the post. This is a research and education competition and I am iterating on its setup.", "votes": null } ], "comments": [ { "id": 3398904, "author_name": "andandand", "author_url": "", "post_date": "01/29/2026 21:58:19", "content": "Thanks for the feedback! Please don't feel obliged to delete the post. This is a research and education competition and I am iterating on its setup.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "3398888": "I just realized the IDs and names of the training data are sorted, my hypothesis was: 'Maybe it is possible to know which jaguar is which based on label proximity.' To my huge disdain, I was correct:\n\n\n\nTo the competition hosts, please consider shuffling the test fold labels or change the numbering strategy to avoid this being an easy way to the top 10.\n\n[here is a link to my tests.](https://www.kaggle.com/code/elcientfico/ordering-leakage-test)\n\nif you deem it adequate I can remove this post and the code proof but is important to acknowledge this kind of data problems to keep the competition fair.", "3398904": "Thanks for the feedback! Please don't feel obliged to delete the post. This is a research and education competition and I am iterating on its setup." }, "source": "meta" }