abeltoolit / hangman-ai Goto Github PK

This project forked from evelynjgao/hangman-ai

A Hangman AI written in Python that accesses a provided API through HTTP posts and requests. It guesses lyrics to rap songs with a win rate of at least 20% for 100 games played. My code hangman.py implements the dictionary that I sourced from the lyrics through playing this game, dict. I began building this dictionary by running the game about 500 times and appending the lyrics to the file in a certain format, namely only alphabetical characters with each word on a new line. I then reset. As the program runs, the dictionary improves as does the win rate. This dictionary includes repeats so that more frequently occuring words are more likely to be guessed. I split the phrase into words and worked on solving the ones with the least unknown characters first. Based off the words in the dictionary, I checked to see which words can fit and the frequency of letters that can occur in these possibilities to make the best educated guess. I also sourced a most frequently appearing letter string from sorting the dictionary and using a counter to get the frequencies of the letter appearences within the dictionary itself. This is the current state of dict after 200 runs since reset of the win rate and games played, with a win rate of exactly 25%.