Napovedovanje dolžin smučarskih skokov

diplomsko delo

Sara Logar (Author), Petar Vračar (Mentor)

Abstract

Smučarski skoki so zelo kompleksen šport, pri katerem majhna variacija nekega dejavnika lahko povzroči velik odklon. V diplomskem delu se ukvarjamo z napovedovanjem dolžin smučarskih skokov s pomočjo metod strojnega učenja. Zbrane podatke o skokih smo obogatili z novimi atributi, pridobljenimi z gručenjem. Za modeliranje podatkov smo uporabili metodo naključnih gozdov. Model smo preizkusili v dveh nalogah: regresijskem napovedovanju dolžin skokov in napovedovanju končne razvrstitve tekmovalcev na koncu sezone. Rezultati so pokazali, da ima model napako, ki bi na sto metrski skakalnici pomenila 6 metrov. Pri končni razvrstitvi se povprečno zmoti za eno mesto, drsenje časovnega okna pa postopoma izboljšuje napovedno zmogljivost. Pričujoče delo prispeva k širitvi uporabe metod strojnega učenja na to kompleksno področje in odpira prostor za nadaljnje raziskave.

Keywords

športna analitika;modeliranje podatkov;smučarski skoki;napovedovanje vrstnega reda;računalništvo;matematika;interdisciplinarni študij;univerzitetni študij;diplomske naloge;

Data

Language:	Slovenian
Year of publishing:	2025
Typology:	2.11 - Undergraduate Thesis
Organization:	UL FRI - Faculty of Computer and Information Science
Publisher:	[S. Logar]
UDC:	004.942:796.925(043.2)
COBISS:	229998339
Views:	136
Downloads:	36
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Prediction of ski jump distances
Secondary abstract:	Ski jumping is a complex sport in which small changes in certain factors can result in significant differences in performance. This thesis focuses on predicting ski jump lengths using machine learning methods. We enriched the collected jump data with new attributes obtained by clustering. We used the random forest method to model the data. We tested the model on two tasks: regression-based prediction of jump lengths and forecasting competitors' final rankings at the end of the season. The results showed that the model has an error of 6 meters on a 100-meter ski jump. The final ranking deviates by one place on average, and sliding the time window gradually improves the predictive performance. The present work contributes to expanding the application of machine learning methods to this complex area and opens up space for further research.
Secondary keywords:	sports analytics;data modeling;ski jumping;ranking prediction;computer science;computer and information science;computer science and mathematics;interdisciplinary studies;diploma;
Type (COBISS):	Bachelor thesis/paper
Study programme:	1000407
Embargo end date (OpenAIRE):	1970-01-01
Thesis comment:	Univ. v Ljubljani, Fak. za računalništvo in informatiko
Pages:	1 spletni vir (1 datoteka PDF (43 str.))
ID:	26010927