ogs-data-analysis-tools

A collection of tools for fetching and analyzing user data from OGS

This repository houses various small utilities (mostly consisting of Python notebooks) for fetching and analyzing user data from the Online Go Server (OGS). Each utility is housed in its own top-level folder. To build, download the entire folder for a given utility. At this time the scope of each utility here is too small to warrant a full separate repository, which is why they are maintained here.

Description of individual utilities

• player_data_tools:
  • Tools for dumping and analyzing individual player data.
  • Dump datasets of all of a player's game information (ranks, outcomes, etc.)
  • Basic cleaning of datasets for use in subsequent statistics/data analysis applications
  • Dump player's games in JSON and SGF formats
    • Games can be pre-filtered using raw and cleaned datasets
  • Use KataGo to analyze full games, including multiple games at once
    • Also produces clean Pandas-ready JSON files with the analysis data
    • A few pre-analyzed professional games are provided. 2000 visits per root.
  • Analytics using above tools
    • Sharpness coefficient: Measure the sharpness of each move