Welcome to Bayesian Workflows using CmdStanPy!

We will be predicting the results of Women's World Cup 2019!

Today's tutorial and dependencies will be provided in Google Colab

This requires a Google account, allowing you to save a copy of the notebook to your personal Google Drive.

Tutorial link: bit.ly/pyladiesstan3

However, if you prefer to work locally on your machine, here are steps to install CmdStanPy and tutorial requirements:

1. Navigate to a desired directory
2. git clone https://github.com/nyc-pyladies/2019-cmdstanpy-bayesian-workshop.git
3. cd 2019-cmdstanpy-bayesian-workshop/
4. python3 -m venv venv
5. source venv/bin/activate
6. pip install -r requirements.txt

In order to use CmdStanPy, we need to install the shell interface to Stan using CmdStan

7. CmdStanPy provides a utility to perform this install. From within Python you can run this utility via the following

import cmdstanpy
cmdstanpy.install_cmdstan()

(This may take a few minutes to install)

Take note of the install directory, as well as the install version. You should see the following logged in the first couple of lines from installation:

Install directory: /Users/(your username)/.cmdstanpy

Downloading CmdStan version 2.20.0

8. Set environment variable for CMDSTAN

export CMDSTAN='/Users/(your username)/.cmdstanpy/cmdstan-2.20.0'

(The path and cmdstan version should match the log output from Step 7)

9. Run jupyter notebook
10. Navigate to tutorial in notebooks/WWC2019.ipynb