\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6832115%2F90ab2a60504a08b8a96bc05e046c814f%2FScreenshot%202024-11-03%20123236.png?generation=1730629980905747&alt=media\")
I've now made public two notebooks and the underlying library for my 2nd place result:
\nPlease leave questions about my code here, rather than in the original solution thread (https://www.kaggle.com/competitions/ariel-data-challenge-2024/discussion/543853), which is quite busy already.
\nI've done my best to structure and comment the code, but it will be quite challenging if you want to study it in detail. I recommend starting with ariel_gp.py->define_prior(), which implements the Bayesian prior as I described in my solution thread. It'll take some time to grasp, but you will see how we can really translate our description of the physics directly into code.
", "messages": [ { "id": 3035224, "postDate": "2024-11-03T06:48:24.593Z", "content": "I've now made public two notebooks and the underlying library for my 2nd place result:
\nPlease leave questions about my code here, rather than in the original solution thread (https://www.kaggle.com/competitions/ariel-data-challenge-2024/discussion/543853), which is quite busy already.
\nI've done my best to structure and comment the code, but it will be quite challenging if you want to study it in detail. I recommend starting with ariel_gp.py->define_prior(), which implements the Bayesian prior as I described in my solution thread. It'll take some time to grasp, but you will see how we can really translate our description of the physics directly into code.
", "rawMarkdown": "I've now made public two notebooks and the underlying library for my 2nd place result:\n- If you want to reproduce my submissions (either the original 2nd place or with later improvements based on suggestions from @cnumber), or make your own: https://www.kaggle.com/code/jeroencottaar/ariel-2nd-place-submission-notebook\n- If you want to visualize the model (similar to the plots in my solution thread): https://www.kaggle.com/code/jeroencottaar/ariel-2nd-place-visualization-notebook\n\nPlease leave questions about my code here, rather than in the original solution thread (https://www.kaggle.com/competitions/ariel-data-challenge-2024/discussion/543853), which is quite busy already.\n\nI've done my best to structure and comment the code, but it will be quite challenging if you want to study it in detail. I recommend starting with ariel_gp.py->define_prior(), which implements the Bayesian prior as I described in my solution thread. It'll take some time to grasp, but you will see how we can really translate our description of the physics directly into code.\n", "votes": 16 }, { "id": 3035332, "postDate": "2024-11-03T09:35:23.470Z", "content": "TY. I'll try to learn that in depth.
\nI want to point out that I'm not sure CC-BY-SA is acceptable. Originally it should be CC-BY-4.0 by the winner license rules:
\n\n\n\n\n
\n- WINNER LICENSE.
\n
\n Under Section 11 (Winners Obligations) of the General Rules below, you hereby grant and will grant Competition Sponsor the following license(s) with respect to your Submission if you are a Competition winner:
\n Open Source: You hereby license and will license your winning Submission and the source code used to generate the Submission under the Open Source Initiative-approved license CC BY 4.0 (see www.opensource.org)
And copyleft is more limiting. However, the host would probably not care since he will not do anything commercial with it anyway. But I ask myself if this code would be useable in other Kaggle comps' since we usually need to use either \"generally commercially available software\" and/or open source code that we can deliver under CC BY 4.0 in order for our solutions not to limit any commercial application. Since copyleft is less permissible and has problems with commercial applications, it may raise problems with reusing it on Kaggle, which is sad IMO.
", "rawMarkdown": "TY. I'll try to learn that in depth. \nI want to point out that I'm not sure CC-BY-SA is acceptable. Originally it should be CC-BY-4.0 by the winner license rules:\n\n>>1. WINNER LICENSE.\nUnder Section 11 (Winners Obligations) of the General Rules below, you hereby grant and will grant Competition Sponsor the following license(s) with respect to your Submission if you are a Competition winner:\nOpen Source: You hereby license and will license your winning Submission and the source code used to generate the Submission under the Open Source Initiative-approved license CC BY 4.0 (see www.opensource.org) \n\nAnd copyleft is more limiting. However, the host would probably not care since he will not do anything commercial with it anyway. But I ask myself if this code would be useable in other Kaggle comps' since we usually need to use either \"generally commercially available software\" and/or open source code that we can deliver under CC BY 4.0 in order for our solutions not to limit any commercial application. Since copyleft is less permissible and has problems with commercial applications, it may raise problems with reusing it on Kaggle, which is sad IMO. ", "votes": 1, "replies": [ { "id": 3035334, "postDate": "2024-11-03T09:41:46.413Z", "content": "I'll see what the organizers requests. I picked CC BY-SA 4.0 to make it consistent with the dataset license (CC BY 4.0 is not an option there).
", "rawMarkdown": "I'll see what the organizers requests. I picked CC BY-SA 4.0 to make it consistent with the dataset license (CC BY 4.0 is not an option there).", "replies": [ { "id": 3035336, "postDate": "2024-11-03T09:48:35.640Z", "content": "Well, the dataset is CC-BY-NC.
\nBut I won't want to use the dataset in future competitions, so I don't care.
\nYour code, on the other handā¦who knows š¤£
\nWell, as I said, the host would probably not care anyway. But it's still relevant to the Kaggle community.
Ah, I meant this dataset containing my code for the CC BY-SA 4.0 license: https://www.kaggle.com/datasets/jeroencottaar/my-ariel-library
\nI can't select CC BY 4.0 there.
", "rawMarkdown": "Ah, I meant this dataset containing my code for the CC BY-SA 4.0 license: https://www.kaggle.com/datasets/jeroencottaar/my-ariel-library\n\nI can't select CC BY 4.0 there." }, { "id": 3035349, "postDate": "2024-11-03T10:10:17.743Z", "content": "Ah, I can just select 'other' as license. All code is now CC BY 4.0.
", "rawMarkdown": "Ah, I can just select 'other' as license. All code is now CC BY 4.0.", "votes": 1 }, { "id": 3035368, "postDate": "2024-11-03T10:33:25.917Z", "content": "Thank you. Do you mean that CC-BY-4.0 does not appear or that you can't select it? In any case, it's strange. It's the fifth option in the dropdown list.
\n
Whoops, was thrown off by the different formattingā¦
", "rawMarkdown": "Whoops, was thrown off by the different formatting..." }, { "id": 3035394, "postDate": "2024-11-03T11:48:56.117Z", "content": "I'll admit, I got confused by the same thing the first time I tried to find CC BY 4.0 haha
", "rawMarkdown": "I'll admit, I got confused by the same thing the first time I tried to find CC BY 4.0 haha" } ] } ] } ], "comments": [ { "id": 3035332, "author_name": "greySnow", "author_url": "", "post_date": "2024-11-03T09:35:23.470000", "content": "TY. I'll try to learn that in depth.
\nI want to point out that I'm not sure CC-BY-SA is acceptable. Originally it should be CC-BY-4.0 by the winner license rules:
\n\n\n\n\n
\n- WINNER LICENSE.
\n
\n Under Section 11 (Winners Obligations) of the General Rules below, you hereby grant and will grant Competition Sponsor the following license(s) with respect to your Submission if you are a Competition winner:
\n Open Source: You hereby license and will license your winning Submission and the source code used to generate the Submission under the Open Source Initiative-approved license CC BY 4.0 (see www.opensource.org)
And copyleft is more limiting. However, the host would probably not care since he will not do anything commercial with it anyway. But I ask myself if this code would be useable in other Kaggle comps' since we usually need to use either \"generally commercially available software\" and/or open source code that we can deliver under CC BY 4.0 in order for our solutions not to limit any commercial application. Since copyleft is less permissible and has problems with commercial applications, it may raise problems with reusing it on Kaggle, which is sad IMO.
", "votes": 1, "replies": [ { "id": 3035334, "author_name": "Jeroen Cottaar", "author_url": "", "post_date": "2024-11-03T09:41:46.413000", "content": "I'll see what the organizers requests. I picked CC BY-SA 4.0 to make it consistent with the dataset license (CC BY 4.0 is not an option there).
", "votes": 0, "replies": [ { "id": 3035336, "author_name": "greySnow", "author_url": "", "post_date": "2024-11-03T09:48:35.640000", "content": "Well, the dataset is CC-BY-NC.
\nBut I won't want to use the dataset in future competitions, so I don't care.
\nYour code, on the other handā¦who knows š¤£
\nWell, as I said, the host would probably not care anyway. But it's still relevant to the Kaggle community.
Ah, I meant this dataset containing my code for the CC BY-SA 4.0 license: https://www.kaggle.com/datasets/jeroencottaar/my-ariel-library
\nI can't select CC BY 4.0 there.
", "votes": 0, "replies": [] }, { "id": 3035349, "author_name": "Jeroen Cottaar", "author_url": "", "post_date": "2024-11-03T10:10:17.743000", "content": "Ah, I can just select 'other' as license. All code is now CC BY 4.0.
", "votes": 1, "replies": [] }, { "id": 3035368, "author_name": "greySnow", "author_url": "", "post_date": "2024-11-03T10:33:25.917000", "content": "Thank you. Do you mean that CC-BY-4.0 does not appear or that you can't select it? In any case, it's strange. It's the fifth option in the dropdown list.
\n
Whoops, was thrown off by the different formattingā¦
", "votes": 0, "replies": [] }, { "id": 3035394, "author_name": "greySnow", "author_url": "", "post_date": "2024-11-03T11:48:56.117000", "content": "I'll admit, I got confused by the same thing the first time I tried to find CC BY 4.0 haha
", "votes": 0, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "3035224": "I've now made public two notebooks and the underlying library for my 2nd place result:\n- If you want to reproduce my submissions (either the original 2nd place or with later improvements based on suggestions from @cnumber), or make your own: https://www.kaggle.com/code/jeroencottaar/ariel-2nd-place-submission-notebook\n- If you want to visualize the model (similar to the plots in my solution thread): https://www.kaggle.com/code/jeroencottaar/ariel-2nd-place-visualization-notebook\n\nPlease leave questions about my code here, rather than in the original solution thread (https://www.kaggle.com/competitions/ariel-data-challenge-2024/discussion/543853), which is quite busy already.\n\nI've done my best to structure and comment the code, but it will be quite challenging if you want to study it in detail. I recommend starting with ariel_gp.py->define_prior(), which implements the Bayesian prior as I described in my solution thread. It'll take some time to grasp, but you will see how we can really translate our description of the physics directly into code.\n", "3035332": "TY. I'll try to learn that in depth. \nI want to point out that I'm not sure CC-BY-SA is acceptable. Originally it should be CC-BY-4.0 by the winner license rules:\n\n>>1. WINNER LICENSE.\nUnder Section 11 (Winners Obligations) of the General Rules below, you hereby grant and will grant Competition Sponsor the following license(s) with respect to your Submission if you are a Competition winner:\nOpen Source: You hereby license and will license your winning Submission and the source code used to generate the Submission under the Open Source Initiative-approved license CC BY 4.0 (see www.opensource.org) \n\nAnd copyleft is more limiting. However, the host would probably not care since he will not do anything commercial with it anyway. But I ask myself if this code would be useable in other Kaggle comps' since we usually need to use either \"generally commercially available software\" and/or open source code that we can deliver under CC BY 4.0 in order for our solutions not to limit any commercial application. Since copyleft is less permissible and has problems with commercial applications, it may raise problems with reusing it on Kaggle, which is sad IMO. " } }