Predicting the results of esports matches

This bachelor's thesis investigates predictions of esports matches using logistic regression. More specifically, in games Counter-Strike: Global Offensive and Counter-Strike 2(CS:GO, CS2). Data analysing and data processing is done with using esport databases. Logistic regression model is generated using R programming language. A method of structured interview was crucial for obtaining information about esport teams and for making predictions. A method of data analysis was used to create descriptive statistics. The prediction model was generated using the logistic regression method. The main data source for this thesis is esport database HLTV. The obtained data is then copied and analysed in Excel with using the function of descriptive statistics. The final dataset is used as input data for the code in RStudio. The code in R is made using external libraries and packages for generating graphics. After the use of statistical functions, the output is a generated logistic regression model. This estimated model is then described using relevant metrics in the field of prediction models. With the completed prediction model, the goal of this thesis has been achieved. The final logistic regression model predicts esports matches with accuracy of 66 %. It was also calculated that the variable containing the difference in world ranking of the teams has the greatest influence on the prediction. In this case, the accuracy of the logistic regression model is not sufficient to reliably estimate esports matches, even though the overall value is satisfactory.