Perfect short reveal: Human-centered machine learning approach in analyzing goalscoring strategies in soccer

Abstract

Machine learning (ML) offers valuable applications in sports science by analyzing complex games and providing insights for coaching and tactics. This study explores differences in shot strategies between male and female soccer players during the 2019 Women’s World Cup and 2022 Men’s World Cup using ML-driven feature importance. The objective insights derived can guide training designs and tactical strategies, enhancing coaching decisions based on data. Results indicate that ‘duration of pressure’ during shooting is a key feature for both genders. However, for male players, ‘position’ and ‘shot body part’ are more influential. ‘Position’ refers to the player’s location during the shot, while ‘shot body part’ indicates whether the shot was taken using the head, left foot, or right foot. In contrast, ‘play pattern’ and ‘location’ (both x and y coordinates of the shot’s starting point) are more significant for female players, making the specific zone on the pitch to be significant too. ‘Play pattern’ describes shots taken as part of specific strategies (e.g., corner kick, free kick). These findings suggest different tactical approaches: for female athletes, coaches should emphasize creating shot opportunities near the goal and focus on play patterns during training. For male athletes, training could focus more on optimizing shot outcomes based on the body part used. This ML-driven analysis highlights the distinct factors influencing goal-scoring moments in male and female soccer, providing data-backed strategies to refine game tactics and training activities

Emaly Vatne

Assistant Sport Scientist at The Ohio State University Department of Athletics and Human Performance Collaborative

My research interests include applied sport science, effects of recovery interventions, and data analysis, visualization, and engineering.