Predicting sports results using machine learning

V diplomskem delu so predstavljeni zgodovina strojnega učenja in pogosto uporabljani algoritmi, opisano je, kako so se algoritmi razvijali in kateri so bili predhodniki sedanjih. Za preizkus uspešnosti izbranih algoritmov v praktičnem delu naloge je bil razvit program, v katerem je preizkušenih nekaj najpogostejših algoritmov strojnega učenja. V ta namen so bili s programom samodejno pridobljeni podatki o tekmah, ekipah in lestvici angleške nogometne lige ter shranjeni v lokalno podatkovno bazo. Namen razvitega programa in uporabljenih algoritmov strojnega učenja je napovedovanje izidov tekem in števila doseženih golov domačega moštva. Točnost napovedi se giblje med 30 in 50 odstotki, za doseganje boljših rezultatov pa bi potrebovali kakovostnejše in obsežnejše podatke.In the thesis we present the history of machine learning and commonly used algorithms. We describe how algorithms were developed and which algorithms were predecessors to the present ones. To test the performance of selected algorithms in the practical part of the thesis, we develop a program in which we try out some of the most common algorithms in machine learning. For this purpose, we automatically acquire information about the matches, teams and rankings of the English Football League, and store them in a local database. The purpose of the developed program and the used machine learning algorithms is to predict the results of the matches and the number of goals scored by the home team. The accuracy of the forecast ranges between 30 and 50 percent. For better and more comprehensive data is needed to achieve better results