Skip to content

About

A Machine Learning based Web Application that can predict the result of Chess Game by rating of both players and the chess opening played.

Resources

Stars

1 star

Watchers

1 watching

Forks

Latest commit

 

History

49 Commits

Folders and files

Repository files navigation

Chess Game Result Prediction

Entertainment is a feature of human behaviour since the earliest of times, countless games and other forms of entertainment have been devised over the centuries, one of the earliest games recorded in relatively recent human history that it is still played worldwide is Chess, or Checkers. As in any other competitive task, humans have been looking for an advantage in competitive chess, for this any tool available to humans were used to try to obtain knowledge, mechanics and mathematics were used to ”beat” chess from the earliest of times.

Motivation

According to IBM’s Shanon , there are around 1043 possible number of board positions, which is a lower bound estimate, thus making it almost impossible to solve the problem using brute force, and also it explain the complexity of chess itself, we think it might be possible that the hardware of modern computers has improved enough to provide a somewhat accurate prediction, using advanced machine learning techniques, and database technology, using our personal computers.

Magnus Carlsen is a Norwegian chess player who faced world chess champion Viswanathan Anand (India) and won the World Chess Championship in 2013 , becoming the second youngest world chess champion at age 22.

This is a set of just over 20,000 games collected from a selection of users on the site Lichess.org.

  • white_rating - strength of a player based on their performance versus other players.
  • black_rating - strength of a player based on their performance versus other players.
  • opening_name - initial moves of a chess game.
  • Performance Evaluation
    • Splitting the dataset by 67 % for training set and 33 % validation set.
  • Training and Validation
    • Logistic Regression with MinMaxScalar gets a higher accuracy score than other classification models.
    • Logistic Regression
  • Performance Results
    • Training Score: 61.5%
    • Validation Score: 63.1%

Demo

Further Improvements

There are lot of things to improve upon

  • Frontend can be made more nicer
  • Need to improve responsiveness for mobile
  • Learning Neural Nets will try to apply soon
  • More data => More Accuracy => More Reliable

About

A Machine Learning based Web Application that can predict the result of Chess Game by rating of both players and the chess opening played.

Resources

Stars

1 star

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages