Folks
\nThis is not a code competition. This means a couple of things.
\nThe last one has bit me on the rear end twice now.
\nI like to fork shared notebooks to local machine and run them. Two different notebooks that I forked had code to erase all files in the working directory - sadly I saved the download to my working directory and twice now I have erased all my work. The first was obvious and I should have seen that one - the second not so clear.
", "messages": [ { "id": "2783984", "postDate": "04/30/2024 03:53:38", "content": "Folks
\nThis is not a code competition. This means a couple of things.
\nThe last one has bit me on the rear end twice now.
\nI like to fork shared notebooks to local machine and run them. Two different notebooks that I forked had code to erase all files in the working directory - sadly I saved the download to my working directory and twice now I have erased all my work. The first was obvious and I should have seen that one - the second not so clear.
", "rawMarkdown": "Folks\n\nThis is not a code competition. This means a couple of things.\n\n1. You can leave the internet ON.\n2. You don't have to add wheels if the internet is on - simple pip install does the trick.\n3. You don't have to erase all files that you created in order for a submission to work.\n4. All work can be done on local machines and the only thing kaggle needs to see is the submission file.\n\nThe last one has bit me on the rear end twice now.\n\nI like to fork shared notebooks to local machine and run them. Two different notebooks that I forked had code to erase all files in the working directory - sadly I saved the download to my working directory and twice now I have erased all my work. The first was obvious and I should have seen that one - the second not so clear.", "votes": null }, { "id": "2787839", "postDate": "05/02/2024 00:49:39", "content": "A cautionary tale! I also like to run code on my own machines, but I avoid forking other people's notebooks. I may scan them for ideas, but I like to write my own code for data preprocessing, model building, and prediction. Code competitions have their upsides, but they make me spend lots of time catering to the requirements of the online environment, so I prefer contests where I can do everything locally except downloading the data and uploading my submissions.
", "rawMarkdown": "A cautionary tale! I also like to run code on my own machines, but I avoid forking other people's notebooks. I may scan them for ideas, but I like to write my own code for data preprocessing, model building, and prediction. Code competitions have their upsides, but they make me spend lots of time catering to the requirements of the online environment, so I prefer contests where I can do everything locally except downloading the data and uploading my submissions.", "votes": null }, { "id": "2790072", "postDate": "05/03/2024 02:43:27", "content": "Been doing kaggle for lots of years and this is the first competition were I got bit by the erasing of files and I have forked a ton of them over the years to local machines. The issue on my part it that I let it happen twice :)
", "rawMarkdown": "Been doing kaggle for lots of years and this is the first competition were I got bit by the erasing of files and I have forked a ton of them over the years to local machines. The issue on my part it that I let it happen twice :)", "votes": null }, { "id": "2790369", "postDate": "05/03/2024 06:52:18", "content": "So we can make predictions locally and submit only the .csv?
", "rawMarkdown": "So we can make predictions locally and submit only the .csv?", "votes": null }, { "id": "2790403", "postDate": "05/03/2024 07:07:26", "content": "Since this is not a \"Code Competition\", the answer is yes, and that's what I've been doing (note that I'm only a competitor and I'm not speaking for the organizers).
", "rawMarkdown": "Since this is not a \"Code Competition\", the answer is yes, and that's what I've been doing (note that I'm only a competitor and I'm not speaking for the organizers).", "votes": null }, { "id": "2828456", "postDate": "05/22/2024 05:43:03", "content": "I have a question.Since only submission file is enough, will there be private score?
\naccording to newly updated metric, there is new library in private set. I am so confused.
There was not change to the test data with the new metric - just a better explanation of its contents. There was also code included to show how the new values are being computed in the discussion of the change.
\nWhen you do a submission both a private and a public score are being computed - they just show you the public score now. So only the scoring has changed - all other items remain as they were.
", "rawMarkdown": "There was not change to the test data with the new metric - just a better explanation of its contents. There was also code included to show how the new values are being computed in the discussion of the change.\n\nWhen you do a submission both a private and a public score are being computed - they just show you the public score now. So only the scoring has changed - all other items remain as they were.", "votes": null } ], "comments": [ { "id": 2787839, "author_name": "dslate", "author_url": "", "post_date": "05/02/2024 00:49:39", "content": "A cautionary tale! I also like to run code on my own machines, but I avoid forking other people's notebooks. I may scan them for ideas, but I like to write my own code for data preprocessing, model building, and prediction. Code competitions have their upsides, but they make me spend lots of time catering to the requirements of the online environment, so I prefer contests where I can do everything locally except downloading the data and uploading my submissions.
", "votes": null, "replies": [ { "id": 2790072, "author_name": "pcjimmmy", "author_url": "", "post_date": "05/03/2024 02:43:27", "content": "Been doing kaggle for lots of years and this is the first competition were I got bit by the erasing of files and I have forked a ton of them over the years to local machines. The issue on my part it that I let it happen twice :)
", "votes": null, "replies": [] } ] }, { "id": 2790369, "author_name": "yuanzhezhou", "author_url": "", "post_date": "05/03/2024 06:52:18", "content": "So we can make predictions locally and submit only the .csv?
", "votes": null, "replies": [ { "id": 2790403, "author_name": "dslate", "author_url": "", "post_date": "05/03/2024 07:07:26", "content": "Since this is not a \"Code Competition\", the answer is yes, and that's what I've been doing (note that I'm only a competitor and I'm not speaking for the organizers).
", "votes": null, "replies": [] } ] }, { "id": 2828456, "author_name": "henryzhaowong", "author_url": "", "post_date": "05/22/2024 05:43:03", "content": "I have a question.Since only submission file is enough, will there be private score?
\naccording to newly updated metric, there is new library in private set. I am so confused.
There was not change to the test data with the new metric - just a better explanation of its contents. There was also code included to show how the new values are being computed in the discussion of the change.
\nWhen you do a submission both a private and a public score are being computed - they just show you the public score now. So only the scoring has changed - all other items remain as they were.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2783984": "Folks\n\nThis is not a code competition. This means a couple of things.\n\n1. You can leave the internet ON.\n2. You don't have to add wheels if the internet is on - simple pip install does the trick.\n3. You don't have to erase all files that you created in order for a submission to work.\n4. All work can be done on local machines and the only thing kaggle needs to see is the submission file.\n\nThe last one has bit me on the rear end twice now.\n\nI like to fork shared notebooks to local machine and run them. Two different notebooks that I forked had code to erase all files in the working directory - sadly I saved the download to my working directory and twice now I have erased all my work. The first was obvious and I should have seen that one - the second not so clear.", "2787839": "A cautionary tale! I also like to run code on my own machines, but I avoid forking other people's notebooks. I may scan them for ideas, but I like to write my own code for data preprocessing, model building, and prediction. Code competitions have their upsides, but they make me spend lots of time catering to the requirements of the online environment, so I prefer contests where I can do everything locally except downloading the data and uploading my submissions.", "2790072": "Been doing kaggle for lots of years and this is the first competition were I got bit by the erasing of files and I have forked a ton of them over the years to local machines. The issue on my part it that I let it happen twice :)", "2790369": "So we can make predictions locally and submit only the .csv?", "2790403": "Since this is not a \"Code Competition\", the answer is yes, and that's what I've been doing (note that I'm only a competitor and I'm not speaking for the organizers).", "2828456": "I have a question.Since only submission file is enough, will there be private score?\naccording to newly updated metric, there is new library in private set. I am so confused.", "2829257": "There was not change to the test data with the new metric - just a better explanation of its contents. There was also code included to show how the new values are being computed in the discussion of the change.\n\nWhen you do a submission both a private and a public score are being computed - they just show you the public score now. So only the scoring has changed - all other items remain as they were." }, "source": "meta" }