<![CDATA[The rapid growth of Mobile Legends: Bang Bang esports produces large volumes of gameplay statistics that remain underutilized beyond individual match summaries. This study compares K-Means and Hierarchical Clustering to group player playstyles using five gameplay variables (kill, death, assist, gold, and match duration) from 990 valid records in the MPL Cambodia Season 6 - BoxMatch dataset, following CRISP-DM preprocessing and Min-Max normalization. The Elbow Method identified K=3 as the optimal number of clusters. K-Means produced three interpretable groups - Carry/Damage Dealer, Late Game Fighter, and Sacrificial/Tank - with a Silhouette Score of 0.252 and a balanced distribution (42.6%, 18.5%, 38.9%), outperforming Hierarchical Clustering with Ward linkage (Silhouette Score 0.2203, less balanced distribution). The results were deployed into an interactive Flask-based web application for dataset upload, clustering visualization, and method comparison, which was validated through black box testing. This research demonstrates that combining clustering with web-based visualization can effectively reveal player playstyle patterns from competitive Mobile Legends gameplay statistics.]]>
Copyrights © 2026
