Usporedba uspješnosti različitih algoritama strojnog učenja na različitim skupovima podataka

undergraduate thesis

• Title (english): Comparing Efficiency of Various Machine Learning Algorithms on Different Datasets
• Author: Ivan Anđelković
• Mentor: Marin Šilić (mentor)
• Committee member: Marin Šilić (predsjednik povjerenstva)
• Committee member: Petar Afrić (član povjerenstva)
• Committee member: Lucija Šikić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2020-07-03, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Algoritmi strojnog učenja znaju biti složeni. U ovom radu prikazan je No Free
Lunch teorem koji naglašava nepostojanje optimalnog algoritma strojnog ucenja. Pri-
kazani su problemi i pozitivne stvari vezane uz nadzirano strojno ucenje. Objašnjeni
su glavni tradicionalni algoritmi klasifikacije. Odabrani su predstavnici tih algoritama
te su stavljeni u uporabu na izabranim skupovima podataka. Rezultati uporabe tablično
prikazani te kasnije komentirani osnovano na ocekivanjima po algoritmima. Prikazan
je jedan način formatiranja podataka za uporabu. Objašnjeni osnovni načini validacije
rezultantnih modela. Prikazane mane točnosti i prikazana F1-mjera kao alternativa u
slučajevima kada je to prikladno.

Abstract (english)

Machine learning algorithms can be complex. In this work the No Free Lunch theorem is shown which emphasizes the fact that an optimal machine learning algorithm
does not exist. The main issues and positives are presented which are mostly tied to
supervised machine learning. Main traditional classification algorithms are shown and
explained. Some of the algorithms are later chosen and put to work on selected datasets. The results of this activity are shown using tables and later discussed based on
the expectations which were in place. This work explains one way in which datasets
are formatted for use. Some of the main ways of model validation are explained. The
flaws of accuracy as a way of validations are shown and the alternative, F1-score is
explained and put to use on datasets where it is appropriate.

• Keywords: strojno učenje; nadzirano strojno učenje; klasifikacija; SVM; KNN; linearna regresija; stabla odluke; naivni bayes klasifikator; neuronske mreže; točnost; F1-mjera; preciznost; odziv; sklearn
• Keywords (english): machine learning; supervised machine learning; classification; SVM; KNN; linear regression; decision trees; naive bayes classificator; neural networks; accuracy; F1-score; precision; recall; sklearn
• Language: croatian
• URN:NBN: urn:nbn:hr:168:080611
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-14 08:18:05