Uporaba podatkovnega rudarjenja v športu

magistrsko delo Organizacija in management informacijskih sistemov

Matjaž Ogrinc (Author), Mirjana Kljajić Borštnar (Mentor)

Abstract

V podatkih se skrivajo potencialno koristne informacije, ki jih lahko izkoristimo za pridobitev novega, uporabnega znanja. V magistrski nalogi smo obravnavali problem analize igre hokeja na ledu z uporabo podatkovnega rudarjenja. Glavni cilj naloge je predstaviti možnosti in uporabnost podatkovnega rudarjenja v športu ter s tem prispevati k razvoju hokeja na ledu doma in po svetu. Po drugi strani pa želimo spodbuditi tudi vodenje statistike, ki je podlaga za uporabo podatkovnega rudarjenja v športu. Rešitev problema smo izvedli po korakih procesa za odkrivanje znanja v podatkih in metodologije CRISP-DM. Na podlagi Evklidske razdalje smo praktično prikazali kako iščemo podobnosti med primeri. Njihovo povezanost pa smo preverili s pomočjo Pearsonovega koeficienta korelacije in postavili homogene napadalne trojke. S programom Orange, smo izdelali modele za uvrščanje igralcev na igralne pozicije (metoda k-NN), razvrstitev ekip in igralcev v skupine (hierarhično razvrščanje, metoda voditeljev) in za vpliv igralca na uspeh ekipe (agoritem CN2, Naive Bayesov klasifikator, odločitvena drevesa in nevronske mreže). Kot pomoč pri vizualizaciji rezultatov smo uporabili ustrezne diagrame. V zaključku smo izdelali SWOT analizo in predstavili statistike, ki bi jih bilo potrebno uvesti za resnejše analize v slovenskem hokeju. Izdelani modeli so v pomoč strokovnemu in učinkovitemu pristopu k hokejski igri ter so z modifikacijami uporabni tudi za druge športe. Koristni so za igralce in trenerje pri analizi igre, moštva in nasprotnikov, managerjem ter vodstvu klubov pri nakupu in menjavi igralcev, iskalcem talentov, sponzorjem, novinarjem…

Keywords

podatkovno rudarjenje;odkrivanje znanja v podatkih;Orange;

Data

Language:	Slovenian
Year of publishing:	2016
Typology:	2.09 - Master's Thesis
Organization:	UM FOV - Faculty of Organizational Sciences
Publisher:	[M. Ogrinc]
UDC:	004.5
COBISS:	7566099
Views:	1319
Downloads:	155
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	USAGE OF DATA MINING IN SPORTS
Secondary abstract:	Potentially useful informations are hiding in data that can be used to obtain a new useful knowledge. In this master's thesis we address the problem of analysis of the ice hockey game with the usage of data mining. The main goal of the project is to present the possibilities and usability of data mining in sports, thereby contributing to the development of ice hockey at home and worldwide. On the other hand, we want to encourage the collecting of statistics, which is the basis for the use of data mining in sports. For solving the problem we have used knowledge discovery in data process step by step and CRISP-DM methodology. Based on measurements of Euclidean distance we practically demonstrated how we search for similarities between the examples. We checked their relationships using Pearsons correlation coefficient and setup homogeneous offensive lines. By usnig Orange Data mining program, we have developed models for classifying players on playing positions (k-NN method), classification of teams and players in clusters (hierarchical clustering, k-Means Clustering) and the player's contribution to team succes (CN2 algorithm, Naive Bayes classification, decission trees and neural networks). Appropriate diagrams were used to assist the visualization of the results. Master's thesis is concluded by a SWOT analysis and introduction of stastistics, which should be used for serious analysis of the Slovenian ice hockey. Constructed models are representing contribution to the professional and efficient approach to hockey game and other sports. They are useful for players and coaches in the analysis of the game, the team and opponents, team managers in the purchase and replacement of players, scouts, sponsors, journalists…
Secondary keywords:	data mining;knowledge discovery data;Orange;ice hockey;
URN:	URN:SI:UM:
Type (COBISS):	Master's thesis/paper
Thesis comment:	Univ. Maribor, Fak. za organizacijske vede
Pages:	89 f.
ID:	9132438