Prediktivna analiza podataka s vremenskim slijedom iz domene sporta

master's thesis

• Title (english): Predictive analysis of time series data in sports
• Author: Nino Črljenec
• Mentor: Marko Banek (mentor)
• Committee member: Marko Banek (predsjednik povjerenstva)
• Committee member: Boris Vrdoljak (član povjerenstva)
• Committee member: Damir Pintar (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2014-07-04, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Cilj ovog rada bio je razviti model dubinske analize podataka za predviđanje ishoda nogometne utakmice na temelju prethodnih rezultata. Željeni ishod obavljenog istraživanja je bio odrediti idealni vremenski interval koji daje najbolje rezultate predviđanja.
Kao pristup rješavanju navedenog problema odabran je CRISP – DM model. Podaci za analizu su preuzeti s web stranice sportskih rezultata te transformirani i učitani u bazu podataka.
Izgrađena su tri različita modela podataka za predviđanje ishoda utakmice s naglaskom na različite vremenske intervale. Svaki od modela evaluiran je pomoću programskog alata Weka koji je potom generirao prediktivne modele pomoću tri različita algoritma: stabla odluke J48, linearne regresije i logističke regresije.
Za svaki model, dani su rezultati predikcije koji su međusobno uspoređeni te je odabran najbolji za korištenje budućih predviđanja. Također, razvijeno je korisničko sučelje za pregledniji prikaz dobivenih rezultata.

Abstract (english)

The topic of the thesis was to develop a data mining model for predicting the outcome of football matches based on previous results. The desired outcome of the research was to determine the ideal time interval that gives the best prediction results.
The CRISP - DM model was selected as a solution for described problems. Data for the analysis was downloaded from the sports results website which was later transformed and loaded into the database.
Three different data models were built, each of them using different time intervals for prediciting the outcome. Each of the models was evaluated using the Weka software tool which then generated predictive models by using three different algorithms: J48 decision tree, linear regression and logistic regression.
For each built model, the results of the predictions were compared with other models. After that the best model was selected for future predictions. Also, a graphic user interface was developed for better preview of collected data.

• Keywords: dubinska analiza podataka; CRISP – DM model; prediktivna analiza; stohastički procesi; Weka; stablo odlučivanja; linearna regresija; logistička regresija; nogometna utakmica
• Keywords (english): data mining; CRISP – DM model; predictive analysis; stochastic process; Weka; decision tree; linear regression; logistic regression; football match
• Language: croatian
• URN:NBN: urn:nbn:hr:168:649677
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 08:18:58