A web interface for exploring PyMC3 traces, inspired by ShinyStan and built with Dash.

Consult the quickstart notebook to start using webmc3.

The simplest way to develop webmc3 is to use the Dockerfile.

First, pull the image from Docker Hub with

docker pull austinrochford/webmc3

or build the image with

docker build -t webmc3 .

Next launch a container using the image.

docker run -d \
    -p $JUPYTER_PORT:8888 \
    -p $DASH_PORT:8050 \
    -v `pwd`:/home/jovyan/webmc3 \
    --name webmc3 webmc3 \
    start-notebook.sh --NotebookApp.token=''

There will not be a Jupyter notebook server available at http://localhost:$JUPYTER_PORT/tree and, once running, the webmc3 app will be available at http://localhost:$DASH_PORT. Most importantly, your fork of webmc3 is on the PYTHONPATH in the container.