This thesis addresses the problem of predicting the outcome of water polo matches using machine learning methods. Compared to other team sports, water polo remains relatively underexplored in the field of sports analytics despite the availability of detailed match statistics. Data about players and matches were collected from the Total Waterpolo platform using web scraping techniques and subsequently processed for analysis. Several machine learning methods were evaluated, including KNN, SVM, decision trees, Random forest, Gradient boost, and neural networks. The performance of these models was compared, based on prediction accuracy, and their suitability for predicting the outcome of water polo matches. Of these methods the decision tree method Random forest proved to be the best.
|
