Finding Feature Importance in Optimized Classification Model: League of Legends Ranked Matches

The unprecedented surge in League of Legends' popularity serves as a compelling phenomenon. This research is aimed at understanding the dynamics of the game and improving team performance, understanding the factors contributing to the results of League of legends matches. We conducted experiments to build classification machine learning algorithms to create a predictive model for League of Legends game outcomes based on specific factors. In the context of esports' rising popularity, this study explores the potential of these algorithms for classification tasks, utilizing labeled data for training and encompassing both training and prediction phases. The research contributes to our understanding of the effectiveness of these machine learning algorithms in predicting esports outcomes and aligns with the industry's anticipated growth in the coming years. KNN outperforms the other algorithms tested in this problem. In conclusion, KNN appears to be the most effective, achieving an accuracy of approximately 73%, with the closest competitor not far behind.