ModuleNotFoundError: No module named 'pandas' when running flaml:latest with Docker

[israel-cj]

When running the following command:

python runbenchmark.py flaml:latest -m docker
I encounter the error:

ModuleNotFoundError: No module named 'pandas'

What I tried:
I attempted to fix this by forcing the installation of pandas in the Dockerfile:

RUN /bench/venv/bin/python3.9 -m pip install --no-cache-dir pandas However, this did not resolve the issue.

I fixed the latest version to 2.3.3 before running the command. The error persists despite manually adding the pandas installation step in the Dockerfile.

Could you please advise on how to resolve this issue or if there’s a missing dependency in the Docker setup?

[PGijsbers]

You can add the requirements in either the frameworks/flaml/requirements.txt file (framework environment) or requirements.txt file (amlb environment) dependening on where the requirement is missing. I cannot tell based on the provided information. I can have a closer look later.

edit: it's strange though... there already is a pandas requirement in both, so this makes little sense. Will need to investigate.

[israel-cj]

I added the requirements.txt with pandas and still I got the same result. I will try a few other things as well

[PGijsbers]

I just did a clean clone and install and it worked fine. Cannot reproduce.

git clone git@github.com:openml/automlbenchmark.git  # or https clone
cd automlbenchmark
python -m venv venv  # use python 3.9
source venv/bin/activate  # ./venv/Scripts/activate.bat on Windows, I believe
python -m pip install -r requirements.txt
python runbenchmark.py flaml:latest -m docker -s force 

Summing up scores for current run:
               id        task  fold framework constraint     result      metric  duration       seed
openml.org/t/3913         kc2     0     flaml       test   0.875541         auc      61.9 1230242422
openml.org/t/3913         kc2     1     flaml       test   0.929654         auc      61.1 1230242423
  openml.org/t/59        iris     0     flaml       test  -0.173929 neg_logloss      61.9 1230242422
  openml.org/t/59        iris     1     flaml       test  -0.057011 neg_logloss      60.6 1230242423
openml.org/t/2295 cholesterol     0     flaml       test -44.932400    neg_rmse      61.9 1230242422
openml.org/t/2295 cholesterol     1     flaml       test -54.191900    neg_rmse      60.6 1230242423

[israel-cj]

Mmm, I see. ok, this is off I think I am doing the same. Thanks for checking anyway

[PGijsbers]

Did you try using -s force?

[israel-cj]

Yes, even I tried an Ubuntu-based system and still same result
I will keep you posted

[israel-cj]

My bad I was following the documentation
git clone https://github.com/openml/automlbenchmark.git --branch stable --depth 1
I should cloned it from the repo. Now, it is working, thank you so much for the help

[PGijsbers]

I'll have a look at updating the docs. Thanks!