Testing
Potential Problems/Research Areas: Possible Database (for first iteration of bot based on elo ratings): https://www.kaggle.com/datasets/datasnaek/chess
Lichess API: https://lichess.org/api#tag/Relations/operation/unfollowUser
Start with a well followed user, find all the followers and the followers of their followers and so on to build a databased of all users. Then loop through the list of these users to get all of their games.
1: Decide if we want create this as a python application or as a website
- Research Jango/React 2: Parser to translate data with moves to translate pulled data into our engine
- Look at already available API 3: The actual ML
- Research feature vectors Possible Backup Plan?
- Create some form of a chess bot on our own
Sprint #1:
Goals: 1: Finish data scrappers and collect data 2: Finish pgn translators (UI) 3: Basic algorithimic AI 4: Experiment with differend ML architectures and try training + evaluating accuracy
Vague Timeline:
By Start of April
- UI and basic engine is done
- Lichess data base created/downloaded
- Architecture for the entire project
- Basic proof of concepts for chess bot o Basic chess bot (basic using the basic chess module) to understand the structure of how a chess bot works o Translate output form this into the UI
Mid April
- Translating the proof of concept into the architecture to implement the Neural Network
Start of May
- Implement the Proof of Concept