This is an ipywidgets implementation of the AnchorViz visualization, see _Chen, Nan-Chen, et al. "AnchorViz: Facilitating classifier error discovery through interactive semantic data exploration"
To install, use pip:
$ pip install ipyanchorviz
For a development installation of the Python library:
$ git clone https://github.com/ORNL/ipyanchorviz.git
$ cd ipyanchorviz
$ pip install -e .
After pip, you need to install node (requires Node.js and Yarn version 1). This will need to be rebuilt when you make a JS change.hen you need to rebuild the JS when you make a code change. The yarn command is run first to install additional needed dependencies.
$ cd js
$ yarn
$ yarn run build
Then to have have the extension work in Jupyter notebook:
$ jupyter nbextension install --py --symlink --overwrite --sys-prefix ipyanchorviz
$ jupyter nbextension enable --py --sys-prefix ipyanchorviz
When actively developing your extension for JupyterLab, run the command:
$ jupyter labextension develop --overwrite ipyanchorviz
You then need to refresh the JupyterLab page when your javascript changes.