Kröckel Pavlina, Bodendorf Freimut
Institute of Information Systems, University of Erlangen-Nuremberg, Nuremberg, Germany.
Front Artif Intell. 2020 Jul 14;3:47. doi: 10.3389/frai.2020.00047. eCollection 2020.
The paper explores process mining and its usefulness for analyzing football event data. We work with professional event data provided by OPTA Sports from the European Championship in 2016. We analyze one game of a favorite team (England) against an underdog team (Iceland). The success of the underdog teams in the Euro 2016 was remarkable, and it is what made the event special. For this reason, it is interesting to compare the performance of a favorite and an underdog team by applying process mining. The goal is to show the options that these types of algorithms and visual analytics offer for the interpretation of event data in football and discuss how the gained insights can support decision makers not only in pre- and post-match analysis but also during live games as well. We show process mining techniques which can be used to gain team or individual player insights by considering the types of actions, the sequence of actions, and the order of player involvement in each sequence. Finally, we also demonstrate the detection of typical or unusual behavior by trace and sequence clustering.
本文探讨了过程挖掘及其在分析足球赛事数据方面的实用性。我们使用了OPTA Sports提供的2016年欧洲杯专业赛事数据。我们分析了一支热门球队（英格兰队）对阵一支弱旅（冰岛队）的一场比赛。2016年欧洲杯上弱旅的成功非常显著，这也使得该赛事别具特色。因此，通过应用过程挖掘来比较热门球队和弱旅的表现很有意思。目的是展示这些类型的算法和可视化分析为足球赛事数据解读提供的选项，并讨论所获得的见解如何不仅能在赛前和赛后分析中，还能在现场比赛期间支持决策者。我们展示了过程挖掘技术，这些技术可通过考虑动作类型、动作顺序以及每个序列中球员参与的顺序来深入了解球队或个体球员。最后，我们还通过轨迹和序列聚类展示了对典型或异常行为的检测。
