Primerjava metod za rangiranje podatkov

Oražem, Urška (author), Virk, Žiga (mentor)

Človek je vedno želel urediti stvari po vrsti glede na pomembnost. To je še posebej pomembno v športu. Želimo uvrstiti ekipe po vrsti od najboljše do najslabše že sredi tekmovanja in tako napovedati rezultat. V tem delu opišemo štiri različne metode razvrščanja ekip na tekmovanjih. Prva metoda uporablja idejo Markovskih verig in se uporablja za razvrščanje spletnih strani pri brskanju s priljubljenim brskalnikom po spletu. Druga temelji na Perron-Frobeniusovem izreku in je zasnovana na uvrščanju ekip fakultet v ZDA v ameriškem nogometu. Tretja temelji na deležu zmag. Uporabljena je za razvrščanje ekip v ameriškem nogometu fakultet v ZDA. Četrta metoda temelji na rešitvi metode najmanjših kvadratov. tekmovanja fakultet v ameriškem nogometu. Vsako metodo smo matematično preučili in jih med seboj primerjali. Našli smo primere, kjer se obnašajo podobno in različno. Z vsako metodo poskušamo napovedati vrstne rede hokejskih ekip v tekmovanju EBEL. Poskušamo napovedati izide glede na število zmag in izgub ter glede na število danih in prejetih golov na posamezni tekmi.

• Language: Slovenian
• Keywords: metode rangiranja, rezultati rangiranja
• Work type: Master's thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2018
• PID: 20.500.12556/RUL-105150
• Publication date in RUL: 30.10.2018

Secondary language

• Language: English
• Title: A Comparison of Methods for Ranking Data

Abstract

A man always wanted to order things in line based on their importance. Especially in sports. We wish to order teams from the best to worst in the middle of the competition and predict results. In this thesis we describe four different methods of ordering teams in competitions. The first described method uses an idea of Markov chains. This method is used in ordering websites while browsing with the popular browser. The second method is based on the Perron-Frobenius theorem and was invented for rating college football teams. The third method is based on ration between wins and loses. It was invented for ranking college football teams. The fourth is based on the least square method. We studied every method and we compared them to each other. We found examples where methods react similarly or differently. We tried to predict the order of hockey teams in EBEL competition and tried to predict outcomes based on a number of winnings and losses based on a number of given goals.

• Keywords: ranking methods, ranking scores