Applying data mining methods for predicting sport event outcomes

Područje primjene dubinske analize podataka danas postaje sve šire pa tako i obuhvaća domenu igara na sreću. U ovom radu dubinska analiza podataka primijenjena je na stvarni, povijesni skup podataka o nogometnim utakmicama i podacima vezanim uz sportsko klađenje na njih. Nad tim skupom provedene su sve faze dubinske analize: od početnog čišćenja i pripreme podataka, preko primjene jedne od metoda dubinske analize podataka, do izrade krajnjeg modela i završne analize. Metoda koja je pobliže opisana i korištena je multinomijalna logistička regresija kao generalizacija obične logističke regresije.In recent years data mining application domains are becoming broader and thus include sports betting and lottery games. In this paper, data mining techniques are applied to the real, historical data set of football games and betting odds. Provided data set went through all of data mining phases, which include: cleaning and preparing data, applying one of data mining methods, creating a model and providing final analysis. Data mining function, which is described and used, is multinomial logistic regression, as generalisation of logistic regression