Predikce výsledků esportových zápasů pomocí logistické regrese

The bachelor thesis deals with the topic of prediction models in esports, more specifically in the context of the video game League of Legends. The aim of the thesis is to construct and estimate logistic regression models to predict the outcome of a League of Legends match based on data available before the match starts. The methods used to accomplish the goal of the thesis are document analysis, data analysis and logistic regression. For processing the data and subsequent training and validation of the prediction models, the R programming language in the RStudio development environment is used. The results of the work are three logistic regression models that predict the outcome of a match based on selected variables. The variables used for prediction include the champions chosen by the team, the side of the map on which the team plays, and statistics representing the strength of the team's current roster of players, such as average gold difference and average damage per minute. The models are also compared to each other and one best performing model is selected.