![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F37166%2F7d6a81bc3e8e5816539d01eb505e6ec3%2F6z91kf.jpg?generation=1667406890250847&alt=media\")
@inversion
\nI got 4.848 by this notebook:
\nhttps://www.kaggle.com/code/ttahara/last-aid-20/
There must be some bugs in evaluation.
\n[Edit]
\nMetric Issue has been resolved:
\nhttps://www.kaggle.com/competitions/otto-recommender-system/discussion/363772
@inversion
\nI got 4.848 by this notebook:
\nhttps://www.kaggle.com/code/ttahara/last-aid-20/
There must be some bugs in evaluation.
\n[Edit]
\nMetric Issue has been resolved:
\nhttps://www.kaggle.com/competitions/otto-recommender-system/discussion/363772
Congratz on crushing the \"perfect submission baseline\" !
", "rawMarkdown": "Congratz on crushing the \"perfect submission baseline\" !", "votes": null }, { "id": "2014544", "postDate": "11/02/2022 16:27:08", "content": "I guess that the positive ones are repeatedly counted when they are dupilicated for many times in your predictionβ¦β¦
", "rawMarkdown": "I guess that the positive ones are repeatedly counted when they are dupilicated for many times in your prediction......", "votes": null }, { "id": "2014555", "postDate": "11/02/2022 16:34:58", "content": "
Thanks!π€£γγγγγ
", "rawMarkdown": "Thanks!π€£", "votes": null }, { "id": "2014561", "postDate": "11/02/2022 16:39:35", "content": "I always love time series-ish competitions, things just have to go wrong one way or another. I know this is probably unrelated to the data but still.
", "rawMarkdown": "I always love time series-ish competitions, things just have to go wrong one way or another. I know this is probably unrelated to the data but still.", "votes": null }, { "id": "2014565", "postDate": "11/02/2022 16:41:50", "content": "Nice work! Looks like I need to dig into where the issue is. Stay tuned!
", "rawMarkdown": "Nice work! Looks like I need to dig into where the issue is. Stay tuned!", "votes": null }, { "id": "2014566", "postDate": "11/02/2022 16:45:15", "content": "you just broke the leaderboard on the first day of the competition haha π
", "rawMarkdown": "you just broke the leaderboard on the first day of the competition haha π", "votes": null }, { "id": "2014569", "postDate": "11/02/2022 16:47:52", "content": "Clearly, the simulation is broken. :-)
", "rawMarkdown": "Clearly, the simulation is broken. :-)", "votes": null }, { "id": "2014798", "postDate": "11/02/2022 21:11:55", "content": "Thanks!
\nIt was lucky to notice this bug on the first day of competition.
", "rawMarkdown": "Thanks!\n\nIt was lucky to notice this bug on the first day of competition.", "votes": null }, { "id": "2014800", "postDate": "11/02/2022 21:15:39", "content": "I think so. \"recall@20\" broke the 1.0, ha ha π€£
", "rawMarkdown": "I think so. \"recall@20\" broke the 1.0, ha ha π€£", "votes": null }, { "id": "2014846", "postDate": "11/02/2022 22:44:28", "content": "Well, isn't that a great thing.!
", "rawMarkdown": "Well, isn't that a great thing.!", "votes": null }, { "id": "2014854", "postDate": "11/02/2022 23:03:05", "content": "There must be a bug in the evaluation, as @ttahara mentions.
\nThat means that any claims with regard to the data made so far (\"the task is too easy\", etc) are not founded. All that we know so far is that something about the calculation of the metric is off π
\nWould be great to have a look at the implementation of the metric once it is all squared out π
", "rawMarkdown": "There must be a bug in the evaluation, as @ttahara mentions.\n\nThat means that any claims with regard to the data made so far (\"the task is too easy\", etc) are not founded. All that we know so far is that something about the calculation of the metric is off π\n\nWould be great to have a look at the implementation of the metric once it is all squared out π", "votes": null }, { "id": "2014950", "postDate": "11/03/2022 01:53:58", "content": "I submitted a result with one last item as the prediction based on the 4.848 kernel and got 0.242, so it might be what you said.
", "rawMarkdown": "I submitted a result with one last item as the prediction based on the 4.848 kernel and got 0.242, so it might be what you said.", "votes": null }, { "id": "2014999", "postDate": "11/03/2022 03:16:37", "content": "Haha, that's funny! It's cool to find the mistake so early
", "rawMarkdown": "Haha, that's funny! It's cool to find the mistake so early", "votes": null }, { "id": "2015087", "postDate": "11/03/2022 04:55:21", "content": "Now bugs have been fixed!
\nhttps://www.kaggle.com/competitions/otto-recommender-system/discussion/363772
I am new on Kaggle but I can understand the pain. In one of the competition I was unable to submit my work due some of these bugs. I encourage you to share the solution for further help @ttahara. Keep rocking and sharing.
", "rawMarkdown": "I am new on Kaggle but I can understand the pain. In one of the competition I was unable to submit my work due some of these bugs. I encourage you to share the solution for further help @ttahara. Keep rocking and sharing.", "votes": null }, { "id": "2036600", "postDate": "11/20/2022 00:41:21", "content": "This is why I never join a competition in the first 2 weeks. Lol.
", "rawMarkdown": "This is why I never join a competition in the first 2 weeks. Lol.", "votes": null } ], "comments": [ { "id": 2014542, "author_name": "theoviel", "author_url": "", "post_date": "11/02/2022 16:26:23", "content": "Congratz on crushing the \"perfect submission baseline\" !
", "votes": null, "replies": [ { "id": 2014556, "author_name": "ttahara", "author_url": "", "post_date": "11/02/2022 16:35:48", "content": "Thanks!π€£γγγγγ
", "votes": null, "replies": [] } ] }, { "id": 2014544, "author_name": "sirius81", "author_url": "", "post_date": "11/02/2022 16:27:08", "content": "I guess that the positive ones are repeatedly counted when they are dupilicated for many times in your predictionβ¦β¦
", "votes": null, "replies": [ { "id": 2014800, "author_name": "ttahara", "author_url": "", "post_date": "11/02/2022 21:15:39", "content": "I think so. \"recall@20\" broke the 1.0, ha ha π€£
", "votes": null, "replies": [] }, { "id": 2014950, "author_name": "brilliantzhang", "author_url": "", "post_date": "11/03/2022 01:53:58", "content": "I submitted a result with one last item as the prediction based on the 4.848 kernel and got 0.242, so it might be what you said.
", "votes": null, "replies": [] } ] }, { "id": 2014555, "author_name": "philippsinger", "author_url": "", "post_date": "11/02/2022 16:34:58", "content": "
I always love time series-ish competitions, things just have to go wrong one way or another. I know this is probably unrelated to the data but still.
", "votes": null, "replies": [] }, { "id": 2014565, "author_name": "inversion", "author_url": "", "post_date": "11/02/2022 16:41:50", "content": "Nice work! Looks like I need to dig into where the issue is. Stay tuned!
", "votes": null, "replies": [ { "id": 2014798, "author_name": "ttahara", "author_url": "", "post_date": "11/02/2022 21:11:55", "content": "Thanks!
\nIt was lucky to notice this bug on the first day of competition.
", "votes": null, "replies": [] } ] }, { "id": 2014566, "author_name": "mrgabrielblins", "author_url": "", "post_date": "11/02/2022 16:45:15", "content": "you just broke the leaderboard on the first day of the competition haha π
", "votes": null, "replies": [ { "id": 2014569, "author_name": "inversion", "author_url": "", "post_date": "11/02/2022 16:47:52", "content": "Clearly, the simulation is broken. :-)
", "votes": null, "replies": [] } ] }, { "id": 2014846, "author_name": "priyanshuprajapatis", "author_url": "", "post_date": "11/02/2022 22:44:28", "content": "Well, isn't that a great thing.!
", "votes": null, "replies": [] }, { "id": 2014854, "author_name": "radek1", "author_url": "", "post_date": "11/02/2022 23:03:05", "content": "There must be a bug in the evaluation, as @ttahara mentions.
\nThat means that any claims with regard to the data made so far (\"the task is too easy\", etc) are not founded. All that we know so far is that something about the calculation of the metric is off π
\nWould be great to have a look at the implementation of the metric once it is all squared out π
", "votes": null, "replies": [] }, { "id": 2014999, "author_name": "samsonfha", "author_url": "", "post_date": "11/03/2022 03:16:37", "content": "Haha, that's funny! It's cool to find the mistake so early
", "votes": null, "replies": [] }, { "id": 2015087, "author_name": "ttahara", "author_url": "", "post_date": "11/03/2022 04:55:21", "content": "Now bugs have been fixed!
\nhttps://www.kaggle.com/competitions/otto-recommender-system/discussion/363772
I am new on Kaggle but I can understand the pain. In one of the competition I was unable to submit my work due some of these bugs. I encourage you to share the solution for further help @ttahara. Keep rocking and sharing.
", "votes": null, "replies": [] }, { "id": 2036600, "author_name": "verracodeguacas", "author_url": "", "post_date": "11/20/2022 00:41:21", "content": "This is why I never join a competition in the first 2 weeks. Lol.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2014516": "inversion \nI got 4.848 by this notebook:\nhttps://www.kaggle.com/code/ttahara/last-aid-20/\n\nThere must be some bugs in evaluation.\n \n \n**[Edit]**\n**Metric Issue has been resolved**:\nhttps://www.kaggle.com/competitions/otto-recommender-system/discussion/363772", "2014542": "Congratz on crushing the \"perfect submission baseline\" !", "2014544": "I guess that the positive ones are repeatedly counted when they are dupilicated for many times in your prediction......", "2014555": "", "2014556": "Thanks!π€£", "2014561": "I always love time series-ish competitions, things just have to go wrong one way or another. I know this is probably unrelated to the data but still.", "2014565": "Nice work! Looks like I need to dig into where the issue is. Stay tuned!", "2014566": "you just broke the leaderboard on the first day of the competition haha π", "2014569": "Clearly, the simulation is broken. :-)", "2014798": "Thanks!\n\nIt was lucky to notice this bug on the first day of competition.", "2014800": "I think so. \"recall@20\" broke the 1.0, ha ha π€£", "2014846": "Well, isn't that a great thing.!", "2014854": "There must be a bug in the evaluation, as @ttahara mentions.\n\nThat means that any claims with regard to the data made so far (\"the task is too easy\", etc) are not founded. All that we know so far is that something about the calculation of the metric is off π\n\nWould be great to have a look at the implementation of the metric once it is all squared out π", "2014950": "I submitted a result with one last item as the prediction based on the 4.848 kernel and got 0.242, so it might be what you said.", "2014999": "Haha, that's funny! It's cool to find the mistake so early", "2015087": "Now bugs have been fixed! \nhttps://www.kaggle.com/competitions/otto-recommender-system/discussion/363772", "2016763": "I am new on Kaggle but I can understand the pain. In one of the competition I was unable to submit my work due some of these bugs. I encourage you to share the solution for further help @ttahara. Keep rocking and sharing.", "2036600": "This is why I never join a competition in the first 2 weeks. Lol." }, "source": "meta" }