Analýza herních metrik a výkonnosti hráčů v Ultimate Frisbee s využitím statistických metod

This bachelor's thesis focuses on the analysis of game metrics and player performance in ultimate frisbee using statistical methods. In the first part, analysis of variance and subsequent post-hoc tests are used to examine the existence of regional differences in team performance indicators across the UFA league. This is followed by a cluster analysis (using the k-means method) to classify players into groups based on playing style and performance. In the final part of the thesis, a logistic regression model is constructed to predict match outcomes based on differences in team statistics from previous games, estimating the probability of a home team victory. The model achieved an accuracy of 73% on the test data and similarly performed well in predicting actual games from the beginning of the 2025 season. The results demonstrate that statistical methods can effectively contribute to a deeper understanding of gameplay dynamics, player evaluation, and sports outcome prediction.