This documents how to create the requirements.txt
for this repository. You only need to be worried about this if you want to update the Jupyter Notebooks using new libraries. For just consuming the notebooks, refer to the README.md which documents how to set up a virtual environment and make use of the existing requirements.txt
.
# create virtual environment
virtualenv env-jupyter2kibana
# activate (macOS)
source env-jupyter2kibana/bin/activate
# activate (Windows)
.\env-jupyter2kibana\Scripts\activate
# install notebook dependencies (example)
pip install pandas altair eland matplotlib numpy
# install kernel for jupyter notebooks
pip install ipykernel
python -m ipykernel install --user --name=env-jupyter2kibana
# create requirements.txt
pip freeze > requirements.txt
#deactivate once done
deactivate
jupyter nbconvert --execute --to html index.ipynb --template classic --output-dir='./docs'
jupyter nbconvert --execute --to html viz-1a-flights-histogram.ipynb --template classic --output-dir='./docs'
jupyter nbconvert --execute --to html viz-1b-flights-splom.ipynb --template classic --output-dir='./docs'
jupyter nbconvert --execute --to html viz-2a-cars-splom.ipynb --template classic --output-dir='./docs'
jupyter nbconvert --execute --to html viz-2b-cars-outlier.ipynb --template classic --output-dir='./docs'
jupyter nbconvert --execute --to html viz-2c-cars-violin.ipynb --template classic --output-dir='./docs'
jupyter nbconvert --execute --to html viz-3a-iris-classification.ipynb --template classic --output-dir='./docs'
jupyter nbconvert --execute --to html viz-4a-anomaly-detection.ipynb --template classic --output-dir='./docs'
jupyter nbconvert --execute --to html viz-4b-anomaly-detection-annotation.ipynb --template classic --output-dir='./docs'
jupyter nbconvert --execute --to html viz-4c-anomaly-detection-split.ipynb --template classic --output-dir='./docs'