Analisis Perbandingan Kinerja Algoritma Gaussian Process Classifier (GPC) dan Xgboost untuk Klasifikasi Hasil Pertandingan Mobile Legend World Championship 6

Abstract

The Mobile Legends: Bang Bang M6 tournament showcases a high level of competition with complex gameplay dynamics and professional player performances. This study aims to analyze match data from the M6 tournament to identify key factors influencing team victories and evaluate player performance based on statistics such as kill, death, assist (KDA), hero selection, and MVP achievements. Two classification algorithms, Gaussian Process Classifier (GPC) and XGBoost, were applied to predict match outcomes using these features. The dataset, collected in CSV format from M6 match results, was processed using Python. The preprocessing stage included data cleaning, categorical feature encoding, and numerical normalization. Model performance was evaluated using accuracy, precision, recall, and F1 Score metrics, with results showing that XGBoost achieved superior performance (F1 Score 0.75) compared to GPC (F1 Score 0.74). Additionally, a visualization system was developed using Streamlit to present insights such as team win rates, hero pick and ban frequencies based on unique match IDs, and top players by stage and role. The analysis revealed that players with consistently high KDA and frequent MVP recognition significantly contributed to team success. This study concludes that XGBoost is more effective for predicting match outcomes in highcomplexity datasets, while GPC is more suitable for balanced data and probabilistic interpretation. These findings are expected to assist esports analysts and professional team coaches in making strategic decisions for competitive Mobile Legends tournaments.