Analiza in primerjava metod za ocenjevanje uspešnosti košarkarjev

LUCI, JERNEJ (author), Štrumbelj, Erik (mentor)

Namen diplomskega dela je analiza statistike košarkarjev in iskanje najboljšega načina za ocenjevanje uspešnosti košarkarja. V prvem delu je opisano zbiranje podatkov s treh različnih spletnih strani: stran Evrolige, stran prve španske lige ACB in stran nemške košarkaške lige BBL. V drugem delu so predstavljeni postopki izračunov štirih različnih ocen uspešnosti: Performance Index Rating (PIR), Player Efficency Rating (PER), Four Factors (FF) in Wins Produced (WP). V zadnjem delu so ocene med seboj primerjane glede na tri različne kriterije: iskanje najboljših igralcev v sezoni, spreminjanje ocen uspešnosti preko sezon in uspešnost napovedovanja zmage. Del naloge je posvečen tudi analizi sprememb ocen košarkarjev, ko ti igrajo v različnih ligah.

• Language: Slovenian
• Keywords: košarka, zbiranje podatkov, statistika, analiza podatkov
• Work type: Bachelor thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2018
• PID: 20.500.12556/RUL-100264
• Publication date in RUL: 19.03.2018

Secondary language

• Language: English
• Title: Analysis and comparison of methods for evaluating basketball player efficiency

Abstract

The aim of the thesis is basketball player analysis and determining the best way to estimate player efficency. The first part is dedicated to collecting data from three different web sites: Euroleague home page, Spanish ACB league home page and German BBL league home page. The second part describes four different metrics: Performance Index Rating (PIR), Player Efficency Rating (PER), Four Factors (FF), Wins Produced (WP) and how to compute them. The final part is dedicated to comparing these metrics on three different criteria: finding the best players in the season, performance rating changes across different seasons and win forecasting. Investigation is made about, how efficency estimates change when basketball players play in different leagues.

• Keywords: basketball, web scraping, statistics, data analysis