Poker Hands Dataset

Simple scripts to extract and browse the IRC Poker Database. Note here only hold'em hands are included. Most (over 95%) hands are dropped either because they're not hold'em or because of data corruption.

Usage

Setting up the environment:

python -m venv .venv
source .venv/bin/activate 
pip install -r requirements.txt

Download the compressed file from University of Alberta website, and extract all hold'em hands data into hands.json:

curl -O http://poker.cs.ualberta.ca/IRC/IRCdata.tgz  # or wget, whichever works on your OS
python3 src/extract.py

There will be 9,478,019 hands extracted in total (after roughly 9 hours of running). Majority of hand data will be skipped due to either invalid game type or corrupted records.

Browse the extracted hands in console:

python3 src/browse.py

References

• IRC Poker Database, Computer Poker Research Group, University of Alberta, 2017.
• Miami Data Science Meetup, 2015.