League of Legends Match Outcome Prediction

Overview

This project analyzes competitive League of Legends matches to predict winners and identify key factors influencing match outcomes. Using logistic regression and other machine learning techniques, we model the relationship between in-game objectives and match results.

Key Findings

• Logistic regression achieves 82% accuracy in predicting match winners
• Key predictors include Baron kills, Dragon kills, and champion kills

Data

The dataset contains 7,620 professional League of Legends matches with 57 variables, including team information, match outcomes, and time-series data on in-game events.

Models

• Logistic Regression (base model)
• Logistic Regression with feature selection
• Regularized Regression (Ridge and LASSO)
• Polynomial Regression
• Random Forest
• Ensemble Learning