The Most Predictable League
This work tries to answer the next question: How predictable football really is?
I found this post: https://labtwentyone.tumblr.com/post/147894684062/why-did-we-stop-loving-football and I'm just updating the data and restructuring it as a cookiecutter data science project.
Results
Until 2016/2017, the Spanish league has been the least competitive league out of the 5 leagues examined. This has changed lately, so now English, German, and Italian leagues seems to be less competitive than Spanish. The French league has almost always remained as the most even competition.
Methodology
The metric which measures effectively unpredictability as a proxy to competitiveness is entropy.
"In Information Theory, Entropy is a measurement for uncertainty for an event’s outcome. In our case, the event is a football match and there are three possible outcomes: a win for the first team, a win for the other team or a draw. If the three outcomes are equally probable, the uncertainty is maximal and so is the entropy (log(3)=1.584963). As the probability of a particular outcome will get closer to one - almost no uncertainty - the match entropy will approach zero."
To get the probability for each team to win in a particular game during a specific season, odds from B365 gambling agency has been used.
Data
This Kaggle's European Soccer database, which spans across 8 years from 2008 to 2016 and includes many teams and matches statistics, has been used to examine the top 5 European leagues and determine which is the most competitive league.
To complete the data until the most recently finished season, matches from 2016/17 to 2019/20 has been downloaded from https://www.football-data.co.uk.
Miscelaneous
Virtual environment
make create_environment- Create virtual environmentsource venv/bin/activate- Activate virtual environmentipython kernel install --user --name=venv- Install virtual environment as the kernel of Jupyter Notebooks
Tricks from "Data Science if Software: Developer #lifehacks for the Jupyter Data Scientist"
"Data Science if Software: Developer..." - Peter Bull l ODSC East 2017
pip install nbdime
# in a git repo,
git-nbdiffdriver config --enable
git diff notebooks/
# paste https://gist.github.com/pjbull/221685a8e03a01baaf1e (bit.ly/py-html-config) into:
vim ~/.jupyter/jupyter_notebook_config.py
cd notebooks
touch .ipynb_saveprogress
# so now, when a .ipynb is edited, a directory is created with plain .py and .html copies
Welcome to MkDocs
For full documentation visit mkdocs.org.
MkDocs commands
mkdocs new [dir-name]- Create a new project.mkdocs serve- Start the live-reloading docs server.mkdocs build- Build the documentation site.mkdocs -h- Print help message and exit.
MkDocs project layout
mkdocs.yml # The configuration file.
docs/
index.md # The documentation homepage.
... # Other markdown pages, images and other files.