Dejan Suzovic
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Dejan Suzovic has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
Carlo Castagna
Pegaso Telematic University
1 shared publication- Del P. Wong
Flinders University
1 shared publication - Hongyou Liu
South China Normal University
1 shared publication Magni Mohr
University of Southern Denmark
1 shared publication- Pengpeng Gou
Hong Kong Baptist University
1 shared publication Shunfang Liu
South China Normal University
1 shared publication
Latest Publications
- 21 July 2026
- Published by SAGE Publications in International Journal of Sports Science & Coaching
Analyzing the impact of tactical and technical indicators on match outcomes in the 2023 FIFA Women's World Cup based on a decision tree model
The rapid professionalization of women's soccer has increased match demands and interest in performance analytics at major tournaments such as the FIFA Women's World Cup. However, there remains a need for transparent and interpretable analyses linking multidimensional technical indicators with match outcome in elite women's soccer. This study analyzed 126 team–match observations from the 2023 FIFA Women's World Cup using publicly available aggregated SofaScore team–match statistics. A multidimensional technical-indicator framework covering attacking, possession,build-up, defensive, goalkeeping, and restart actions was established. Match tempo was quantified using Event Pace (technical events per minute of effective playing time), and matches were classified into fast- and slow-tempo groups using a median split. Correlation analysis, independent-samples t-tests, and an interpretable decision-tree model with ROC evaluation were applied. The tournament showed a heterogeneous tempo profile (2.05 ± 0.36 events·min −1 ). Fast-tempo matches involved more long passes, outside-box shots, successful crosses, and ground duels, whereas slow-tempo matches showed more aerial duels won and goalkeeper saves. The decision tree identified shots on target, dribbles won, possession, big chances, aerial duels won, and expected goals as key threshold-based split variables associated with winning versus not winning. The model showed high apparent discrimination (AUC = 0.969), but 10-fold cross-validation indicated more moderate performance (AUC = 0.703). The model should therefore be interpreted as a post-match explanatory classification model rather than a pre-match forecasting tool.
- 21 July 2026
- Published by SAGE Publications in International Journal of Sports Science & Coaching
Analyzing the impact of tactical and technical indicators on match outcomes in the 2023 FIFA Women's World Cup based on a decision tree model
The rapid professionalization of women's soccer has increased match demands and interest in performance analytics at major tournaments such as the FIFA Women's World Cup. However, there remains a need for transparent and interpretable analyses linking multidimensional technical indicators with match outcome in elite women's soccer. This study analyzed 126 team–match observations from the 2023 FIFA Women's World Cup using publicly available aggregated SofaScore team–match statistics. A multidimensional technical-indicator framework covering attacking, possession,build-up, defensive, goalkeeping, and restart actions was established. Match tempo was quantified using Event Pace (technical events per minute of effective playing time), and matches were classified into fast- and slow-tempo groups using a median split. Correlation analysis, independent-samples t-tests, and an interpretable decision-tree model with ROC evaluation were applied. The tournament showed a heterogeneous tempo profile (2.05 ± 0.36 events·min −1 ). Fast-tempo matches involved more long passes, outside-box shots, successful crosses, and ground duels, whereas slow-tempo matches showed more aerial duels won and goalkeeper saves. The decision tree identified shots on target, dribbles won, possession, big chances, aerial duels won, and expected goals as key threshold-based split variables associated with winning versus not winning. The model showed high apparent discrimination (AUC = 0.969), but 10-fold cross-validation indicated more moderate performance (AUC = 0.703). The model should therefore be interpreted as a post-match explanatory classification model rather than a pre-match forecasting tool.