As of today, the version of sklearn printed out in my Kaggle kernel is 0.21.3. I want to use some features that started in the latest stable version 0.22, released 03 Dec 2019. The code !pip install -U scikit-learn==0.22 failed, possibly because no form of internet access is allowed in this competition's code. I hope that kernel competitions are not backdated to stable package versions that were only available at the start of the competition. Has anyone else got this to work? Thanks in advance.
As of today, the version of sklearn printed out in my Kaggle kernel is 0.21.3. I want to use some features that started in the latest stable version 0.22, released 03 Dec 2019. The code !pip install -U scikit-learn==0.22 failed, possibly because no form of internet access is allowed in this competition's code. I hope that kernel competitions are not backdated to stable package versions that were only available at the start of the competition. Has anyone else got this to work? Thanks in advance.
I haven't tried sklearn==0.22, but it might be worth checking your kernel settings. The default option is \"Original\" for docker image but you can change it to \"Latest Available\". I'm not sure if the latest available contains sklearn==0.22 but it is more likely to contain it at least.
", "rawMarkdown": "I haven't tried sklearn==0.22, but it might be worth checking your kernel settings. The default option is \"Original\" for docker image but you can change it to \"Latest Available\". I'm not sure if the latest available contains sklearn==0.22 but it is more likely to contain it at least.", "votes": null }, { "id": "698063", "postDate": "12/18/2019 19:03:11", "content": "@datahobbit : I appreciate the kind reply. Unfortunately, as of today (a day after the close of this competition), \"Latest Available\" in the right panel's Settings still produces sklearn==0.21.3 from Kaggle's point-of-view. Hopefully a system administrator reads this thread! Well, I did get sklearn==0.22 to work in my Kaggle kernel only after I set \"Internet=On\" and did the above pip install, but this was forbidden during the competition. Maybe this'll be allowed for the Late Submissions so that I can see how my new approach would have fared!
", "rawMarkdown": "datahobbit : I appreciate the kind reply. Unfortunately, as of today (a day after the close of this competition), \"Latest Available\" in the right panel's Settings still produces sklearn==0.21.3 from Kaggle's point-of-view. Hopefully a system administrator reads this thread! Well, I did get sklearn==0.22 to work in my Kaggle kernel only after I set \"Internet=On\" and did the above pip install, but this was forbidden during the competition. Maybe this'll be allowed for the Late Submissions so that I can see how my new approach would have fared!", "votes": null } ], "comments": [ { "id": 697428, "author_name": "datahobbit", "author_url": "", "post_date": "12/17/2019 22:39:53", "content": "I haven't tried sklearn==0.22, but it might be worth checking your kernel settings. The default option is \"Original\" for docker image but you can change it to \"Latest Available\". I'm not sure if the latest available contains sklearn==0.22 but it is more likely to contain it at least.
", "votes": null, "replies": [] }, { "id": 698063, "author_name": "qayumkhan", "author_url": "", "post_date": "12/18/2019 19:03:11", "content": "@datahobbit : I appreciate the kind reply. Unfortunately, as of today (a day after the close of this competition), \"Latest Available\" in the right panel's Settings still produces sklearn==0.21.3 from Kaggle's point-of-view. Hopefully a system administrator reads this thread! Well, I did get sklearn==0.22 to work in my Kaggle kernel only after I set \"Internet=On\" and did the above pip install, but this was forbidden during the competition. Maybe this'll be allowed for the Late Submissions so that I can see how my new approach would have fared!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "697362": "As of today, the version of sklearn printed out in my Kaggle kernel is *0.21.3*. I want to use some features that started in the latest stable version **0.22**, released 03 Dec 2019. The code `!pip install -U scikit-learn==0.22` failed, possibly because no form of internet access is allowed in this competition's code. I hope that kernel competitions are not backdated to stable package versions that were only available at the start of the competition. Has anyone else got this to work? Thanks in advance.", "697428": "I haven't tried sklearn==0.22, but it might be worth checking your kernel settings. The default option is \"Original\" for docker image but you can change it to \"Latest Available\". I'm not sure if the latest available contains sklearn==0.22 but it is more likely to contain it at least.", "698063": "datahobbit : I appreciate the kind reply. Unfortunately, as of today (a day after the close of this competition), \"Latest Available\" in the right panel's Settings still produces sklearn==0.21.3 from Kaggle's point-of-view. Hopefully a system administrator reads this thread! Well, I did get sklearn==0.22 to work in my Kaggle kernel only after I set \"Internet=On\" and did the above pip install, but this was forbidden during the competition. Maybe this'll be allowed for the Late Submissions so that I can see how my new approach would have fared!" }, "source": "meta" }