This machine learning project leverages gameweek-specific data from Fantasy Premier League, considering fundamental player statistics such as bonus points, expected points, goals scored, assists, ict_index, influence, threat, and creativity. Its objective is to develop a model that accurately predicts a player’s actual weekly point total in that specific week of the Fantasy Premier League. The six models I developed were Linear Regression, Decision Tree, Random Forest, Boosted Trees, Lasso Regression, and Ridge Regression. Upon thorough evaluation using three key metrics: RMSE, MAE, and R-Squared, the Random Forest model emerged as the greatest model, distinguished by its commendable accuracy yielding an RMSE value of 0.519.
Links: Project Report and GitHub repo
Citation
@online{trujillo2022,
author = {Trujillo, Piero},
title = {Predicting {Points} in {Fantasy} {Premier} {League}},
date = {2022-12-11},
url = {https://suppiero.github.io/projects/fantasy_premier_league/},
langid = {en}
}