Unable to install 3rd party python packages on Dataiku jupyter notebook
I just started exploring Dataiku and wanted to install `ipympl` on the jupyter notebook. But it is not working.
Seems, the notebook is using virtual environment `
/opt/dataiku/pyenv
I tried installing it like--
But you can see there is a permission error, so I tried --
Which also resulted in an error, I have experience on Jupyter notebook on IBM Watson studio but never faced issues like this, can anyone guide me on how to do that, I also did not find any good documentation, I tried to follow this doc but can't find where the DATA_DIR is since `! echo $DATA_DIR` gives null string. Regarding recommended method, I can't find the administration tab anywhere on the Dataiku portal. Please guide me in the right direction. Thanks.
Best Answer
-
Sergey Dataiker, Dataiku DSS Core Designer, Dataiku DSS & SQL, Dataiku DSS Core Concepts, Registered Posts: 365 Dataiker
Dataiku Online indeed has no possibility to create code envs (no Administration tab). You will need to install DSS as VM locally to have that possibility.
Answers
-
Sergey Dataiker, Dataiku DSS Core Designer, Dataiku DSS & SQL, Dataiku DSS Core Concepts, Registered Posts: 365 Dataiker
Hi @epsi95
Yes, by default Jupyter notebooks use built-in code env (DATA_DIR/pyenv). If you need to install python packages, the recommended way is to create a dedicated virtual code env and install packages there:
https://doc.dataiku.com/dss/latest/code-envs/operations-python.html
For the administration page click on the "apps" icon:
When you create a code env and install the required packages, you can change the kernel in the notebook to this new code env:
Hope you will find this information useful.
-
Hi @sergeyd
,Oh okay, thanks for the information. But I can't see the administration tab. I am using 14 days trial though.
-
CoreyS Dataiker Alumni, Dataiku DSS Core Designer, Dataiku DSS Core Concepts, Registered Posts: 1,150 ✭✭✭✭✭✭✭✭✭
Hi @epsi95
and welcome to the Dataiku Community. In addition to Sergey's solution (thank you for accepting it as a solution btw) here is a resource you may find helpful if you haven't accessed it already: Work With Python on Dataiku Online in our Knowledge Base.I hope this helps!