From Data With Love

This is a collection of code and utilities that I have developed whilst playing with data without sufficient adult supervision.

Companion code to my tech blog: http://fromdatawithlove.thegovans.us/

Requirements:

Old Faithful

To install within a virtualenv: pip install -r oldfaithful_requirements.txt

• numpy
• matplotlib
• scikit-learn
• nose (for unit tests)

Chess Social

To install within a virtualenv: pip install -r chess_requirements.txt

• numpy
• matplotlib
• scipy
• networkx
• nose (for unit tests)
• mock (for unit tests)

To Run the Code:

Old Faithful

python old_faithful.py ../data/faithful.csv

There are also other command line options:

• --iterations: Number of iterations for the Gibbs sampler (default 500)
• --save_diagnostics: Whether to save the diagnostic images (default False)
• --output_dir: The directory to save the images to (default '.')
• --burnin: The number of burnin iterations (default 0)

Chess Social

To download the TWIC chess dataset:

python twic_scrape.py

To run using the downloaded TWIC chess dataset:

python run_community_detection.py path/to/twic_chess_data.pgn

There are also other command line options:

• --iterations: Number of iterations for the Gibbs sampler (default 100)
• --output_dir: The directory to save the images to (default '.')
• --burnin: The number of burnin iterations (default 0)
• --min_elo: The minimum elo rating for players to be included (default 2500)
• --p_in: The initial value for the 'IN' edge probabilities (default 0.8)
• --p_out: The initial value for the 'OUT' edge probabilities (default 0.2)