Mobile Legends Win Rate Prediction and Team Recommendation Using Switched Hero Roles

Introduction:Mobile Legends Bang Bang (MLBB) falls under the category of a multi-linebattle arena game whichrequiresplayers to have strong skills and strategic gameplay; team composition is an important factor influencing the chances of winning the game.Novelty Statement:Although there is data currently available for MLBB, two aspects of this game that remain unexplored include: i) win rate prediction using nontraditional roles in heroes, and ii) team composition with switched hero roles. Material and Method:This research aims to address this issueby predicting the win rate of heroes with switched roles. This unpredictability will lead to the formation of a team that can have a significant advantage over the enemy team thus leading to victory. The dataset for this study was formulatedfocusing on 67 heroes in the game. The win rates were generated with real-time simulations where the ally team members remained unchanged to avoid biased results. Result and Discussion: The research utilized two model-building approaches and win rate predictions were made using 12 regression algorithms under 5 feature selection settings. The results showthat LightGBM with AdaBoost as the base estimator provides better results and wasused to formulate 5 teams. A recommendation system was designed to optimizeteam composition from the win rate prediction analysis. To validate the results, we simulated 50 matches with each teamresultingin a 94% win rate.Concluding Remarks: Theresearch exploresswitched hero rolesand provides promising results tohelpteam formation with an increased chance of victorywhen usingnon-traditional hero roles.