In this repository, I store code created during my effort to get better at working with data...and perhaps it will help me to learn something about chess, too.
There is content in the folder pawn_move_evaluation: the Python code therein processeses game data from the Lichess database in order to assess the average accuracy of a given pawn move and subsequently plots it. The data is loaded from the data folder (which is entry in this Git repository). The results of this analysis has been discussed in a Lichess blog post.
I have also started looking into the possibility of predicting the winner of a game just based on the given board state with the use of deep learning in the folder win_predictor. Same Lichess database is used.