Sign up to take part
Registered users can ask their own questions, contribute to discussions, and be part of the Community!
I have a code environment with a bunch of python libraries installed. I'm using this for a data science project.
I'm now trying to run a survival analysis and survival model, with the exploration and model prototype happening in a python notebook. I have tried to install python libraries such as pysurvival, scikit-survival, and lifelines. I am trying one installation at a time, and have encountered multiple issues.
With pysurvival and scikit-survival it's about errors related to dependencies that I haven't been able to solve even by choosing versions compatible with my python version and other libraries versions 😞
With lifelines I'm fortunately not having any dependency problem but I'm encountering the following error (I'm also attaching a screenshot):
Something went wrong during feature update: AxiosError: Network Error
When I remove lifelines from the requirements text, I can save and re-run the code environment successfully. But if I add it back, the error occurs again. This is happening with all lifelines package versions (<= 0.26.4) I've tried. I haven't tried more recent versions though because they require >= python 3.7 and I'm using python 3.6.
Until this time, I had never had any problems installing packages for other data science and NLP tasks in the same code environment.
I would really appreciate any help if anyone might know why this issue might be happening and how to solve it.
Operating system used: Mac OS
Hi @DarioDiaz ,
The "AxiosError" does not always mean the code env installation failed it may have such reach a network timeout.
If the list of package/ pip resolution takes a long time.
To troubleshoot this please open a support chat with Dataiku Support :
Please enable remote support to your instance and we will verify.
With regards to your dependancy errors I will strongly suggest you try with a Python version newer than 3.6. Dataiku supports up to Python 3.11 for code environments and adding another Python version to your installation should be straightforward, assuming you have the right access. Also I wouldn’t recommend you request packages with specific versions (ie pinned packages). The best approach is to let pip decide what versions are compatible between all the packages you have. Once all packages are installed and working you can however “pin” the requested packages to the versions you got installed to prevent further changes.