Analysis of the offensive playing style based on pass event data in the 2023 FIFA Women’s World Cup: an unsupervised machine learning approach

Abstract

The analysis of technical-tactical performance in women’s football has begun to develop exhaustively in recent years, and it must be further identified in the years to come. The objective of this study was to develop and train a segmentation algorithm capable of classifying pass-type event data based on technical-tactical indicators, and to interpret the trends and playing styles of national teams in the FIFA Women’s World Cup 2023 through the clustering and data visualisation techniques. A cross-sectional descriptive design was used to collect 227,393 observations from the 64 matches played in the FIFA Women’s World Cup 2023. The passes were segmented into 5 groups using a K-means algorithm and interpreted using a multivariate descriptive decision tree analysis. The results were visualised through frequency distributions and specific graphics displaying the start and end coordinates of passes on the field. The findings revealed differences in the types of passes executed by the top- and lower-performing teams in the tournament, facilitating the identification of collective play patterns and styles through visual tools. This procedure could be valuable in football for evaluating the performance of one’s own team and opponents, as well as for gaining insights into the playing style of a specific player.