The purpose of this project is to showcase the usage of the Mapper algorithm provided by the Giotto Python library. Have a look at our blog post for more information.
Spin up a conda virtual environment and install the required libraries:
conda create --name mapper python=3.7 && conda activate mapper
conda install jupyter -y
pip install -r requirements.txt
To see the Plotly graphs in JupyterLab, some extra steps are required:
# Avoid "JavaScript heap out of memory" errors during extension installation
# (OS X/Linux)
export NODE_OPTIONS=--max-old-space-size=4096
# (Windows)
set NODE_OPTIONS=--max-old-space-size=4096
# Jupyter widgets extension
jupyter labextension install @jupyter-widgets/jupyterlab-manager@1.1 --no-build
# FigureWidget support
jupyter labextension install plotlywidget@1.4.0 --no-build
# and jupyterlab renderer support
jupyter labextension install jupyterlab-plotly@1.4.0 --no-build
# Build extensions (must be done to activate extensions since --no-build is used above)
jupyter lab build
# Unset NODE_OPTIONS environment variable
# (OS X/Linux)
unset NODE_OPTIONS
# (Windows)
set NODE_OPTIONS=
After running these steps, deactivate and reactivate your conda environment before spinning up JupyterLab.
We applied the Mapper algorithm on a dataset containing 20 000 3-dimensional observations sampled from a Santa Claus shape.
The dataset has been sampled from a mesh object available here through the usage of CloudCompare.
All the analyses are performed in the notebook "Christmas Mapper.ipynb". We briefly describe the Mapper algorithm and we show how the output graph changes by varying:
Enjoy!
In order to run the notebook, the following python packages are required: