Which are the most frequently used tiles on a chess board? Which tile is black most likely to move a pawn onto? What tile are both players most likely to lose a piece on? This repo provides an analysis of >1 million chess moves to answer these questions and more.
Lichess is a free and open-source chess server with an API that provides access to all games played on the site. I use the Berserk Python wrapper to download the top X games from the top Y players on Lichess, resulting in Z total chess moves.
- Create a Lichess account and create a token. Save this token as
lichess.tokenwithin the repository. process.pyfetches and cleans the data from Lichess. Modify the number of top players and the number of games per player at the bottom of the script. The processing script creates a dataframe saved asplays.featherwhere each row is one chess move. The data has the following datatypes:
| column name | data type | description |
|---|---|---|
| white | bool | Is this a move by white. |
| piece | str | 'K', 'Q', 'B', 'N', 'R', or 'P' |
| posx | int | x position [0, 8] |
| posy | int | y position [0, 8] |
| kill | bool | Did this move remove a piece |
| check | bool | Did this move result in check |
| mate | bool | Did this move result in mate |
- The
plot.pyscript has some helpful plotting scripts for creating heat plots to show the frequency that each position on the board is moved to.
