README.md

Code for: "A point-based Bayesian hierarchical model for predicting the outcome of tennis matches"

Requirements

1. Python 2
2. The requirements in requirements.txt.

Getting started

1. Install the requirements listed in requirements.txt. You should be able to do this using pip install -r requirements.txt. Note: On some systems, old versions of gcc may cause Stan to fail to compile the model. In this case, we recommend the conda package manager, which installs the required dependencies. To use conda instead of pip, install the requirements using conda install --file requirements.txt.
2. An example of how to run the model can be found in Example.ipynb.

Overview of the repository

The repository contains the following files:

• Example.ipynb, containing an example of how to run the model
• bayes_point_model.py, the python code implementing the model
• dataset.csv, the data used to fit the model (ATP match results from 2011 onwards)
• requirements.txt, listing the requirements to install the model
• stan_model.stan, the Stan code to fit the model
• winning_prob.py, code calculating the iid win probability given serve-winning probabilities, written inpython.