Apriori and FP-Growth Comparative Analysis of MPL Indonesia Season 13 Hero Drafts

Abstrak

Hero combination selection or drafting is a crucial factor in determining victory in Mobile Legends: Bang Bang (MLBB) games at the professional level such as MPL Indonesia Season 13. However, counter-pick strategies are often based solely on the subjective intuition of players or coaches. This study aims to provide an objective basis for determining winning hero combination patterns by applying the Association Rule Mining (ARM) technique. Two main algorithms, namely Apriori and Frequent Pattern Growth (FP-Growth), are compared to evaluate the performance efficiency and accuracy of the resulting rules. The research data includes 183 winning transactions during the regular season of MPL ID Season 13, with parameters of minimum support 0.05 (5%), minimum confidence 0.40 (40%), and minimum lift 1.2. The results show that the strongest association rules are found in the combinations {Lapu-lapu} → {Fredrinn} (confidence 0.71) and {Cici} → {Fredrinn} (confidence 0.59). In terms of technical performance, the Apriori algorithm recorded a faster execution time than FP-Growth on this dataset. This study concluded that both algorithms produce identical association rule outputs, while Apriori demonstrated faster execution on this small-scale dataset, a finding attributed to the limited transaction volume rather than a universal superiority of Apriori over FP-Growth. The resulting rules can serve as a data-driven strategic recommendation system for professional esports teams in the pick and ban phase.