In this project, we aim to predict rainfall in the African Sahel region with various machine learning and deep learning methods. As data, we use a CICMoD - a climate index collection based on model data from two state-of-the-art earth system models. For details, see: https://github.com/MarcoLandtHayen/climate_index_collection
For now, we're developing in a Docker container with JupyterLab environment, Tensorflow and several extensions, based on martinclaus/py-da-stack.
To start a JupyterLab within this container, run
$ docker pull mlandthayen/py-da-tf:shap
$ docker run -p 8888:8888 --rm -it -v $PWD:/work -w /work mlandthayen/py-da-tf:shap jupyter lab --ip=0.0.0.0
and open the URL starting on http://127.0.0.1...
.
Then, open a Terminal within JupyterLab and run
$ python -m pip install -e .
to have a local editable installation of the package.
Additionally, there's a container image having predict_sahel_rainfall as pre-installed Python package: https://hub.docker.com/r/mlandthayen/predict_sahel_rainfall.
You can use it wherever Docker is installed by running:
$ docker pull mlandthayen/predict_sahel_rainfall:<tag>
$ docker run -p 8888:8888 --rm -it -v $PWD:/work -w /work mlandthayen/predict_sahel_rainfall:<tag> jupyter lab --ip=0.0.0.0
and open the URL starting on http://127.0.0.1...
.
Here, <tag>
can either be latest
or a more specific tag.
You can use it wherever Singularity is installed by essentially running:
$ singularity pull --disable-cache <target.sif> docker://mlandthayen/predict_sahel_rainfall:<tag>
$ singularity run --bind $WORK <target.sif> jupyter lab --no-browser --ip $(hostname) $WORK
Here, <tag>
can either be latest
or a more specific tag.
And <target.sif>
specifies the target file to store the container image.
Project based on the cookiecutter science project template.