Depreciation of python packages and .yml issues

[JelleLagerweij]

Dear developer,
I am preparing for the MolSim2024 course in Amsterdam. For this, I needed to:

• install wsl (and a lot of compiler options with VScode)
• create a conda env for OPS
• create a conda env for ml_molsim

I completed all tasks successfully, however, I did have some slight setbacks when creating the ml_molsim conda environment, which I will go into below.

Foremost, when executing conda env create -f environment.yml -n ml_molsim command, the installation crashed. It surprised me because I was using a clean wsl with a clean conda install. I forgot to document and save the error report (it would have made creating this issue much cleaner), but I solved it by looking at a similar issue online: DeepLabCut/DeepLabCut#2322. In the end, it had something to do with the nb_conda package and how it works together with current notebook versions. I adjusted this in the environment.yml by limiting the version of the notebook package (see below). After this, the installment procedure worked without issues.

An additional issue occurred only after I tried to run the molsim_ml.ipynb file. It crashed in the first relevant code block (under Import package we need). Again, I forgot to save my own error output, however this post solved my issues: https://pypi.org/project/pandas-profiling/. If I understand correctly, the package pandas-profiling is renamed to ydata-profiling. I adjusted this in the environment.yml as well (see below). This would however also need changes in the jupyter notebook and python files. I think that all issues would be resolved if the original environment.yml file would contain version numbers. This would at least make the environment more robust.

Renewed environment.yml file

channels:
  - conda-forge
  - defaults
dependencies:
  - pip
  - ydata-profiling
  - ipywidgets
  - hyperopt
  - kaggle
  - plotly
  - scikit-learn
  - scipy
  - numpy
  - python
  - py-xgboost
  - jupyterlab
  - nb_conda
  - notebook<7.0.0
  - bokeh
  - pandas
  - holoviews
  - seaborn

[kjappelbaum]

thanks for your feedback! We're currently in the process of revising it for this years' school and will take your report into account

[kjappelbaum]

thanks for the feedback, updated in the new default branch.