Chess Tactic Classifier

A program that classifies chess tactics based on the concepts they demonstrate.

Table of Contents

• Abstract
• Timeline
• Data sources
• Technologies
• Club resources
• End Goal

Abstract

To improve their skills, many chess players practice simple chess puzzles of one or two moves. These puzzles can be grouped based on the different concepts or strategies they illustrate, and learning and recognizing these patterns is a key skill for competitive chess players. We are going to build a program that recognizes three types of tactics: forks, skewers and trapped pieces.

Trapped Piece

Fork

Skewer

Timeline

Week 3

• Plan web scraping process
• scrape data from Chess.com
• Get data into .csv files for each tactic type

Example:

• fork.csv
  • FEN, Qe3, d4, white
  • FEN, b6, Be4, black

Week 4

• Clean data, data features
• Potential features
  • squares attacked by piece that moved
  • occupied squares
  • pieces on the same line/diagonal as the piece that moved
• Organize Github repo
• Establish git workflow

Week 5

• Choose and extract features / Design models

Week 6

• figured out how virtual environments work
  • save reqs with $ pip freeze > requirements.txt
  • use $ pip install -r requirements.txt
• Design models, UI design -Adam: attacked piece is defended

Week 7

• Design models, UI design
• 30 each: 1' 34" @ Starbucks
• 60, 30, 30: 2' 2" @ Starbucks
• 60 each: 3' 49" @ Starbucks, error in getting skewers

Week 8

• Polishing project

Week 9

• Buffer week

Data sources

• Chess.com

Technologies

• python for scraping data
• python for machine learning
• python (django) / js for webapp building

Club resources required

• guidance on web scraping and building the webapp

End Goal

A webapp that users can copy and paste a chess position (in the form of an FEN string) and see the position displayed, with the category of tactic the program outputs.