Usporedba postupaka dubinske analize primjenjenih nad biološkim podacima

scientific master's theses

• Title (english): Comparison of Data Mining Techniques Applied on Biological Data
• Author: Mirjana Domazet Lošo
• Mentor: Mirta Baranović (mentor)
• Committee member: Damir Kalpić (predsjednik povjerenstva)
• Committee member: Mirta Baranović (član povjerenstva)
• Committee member: Mladen Varga (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Applied Computing); Zagreb
• Defense date and country: 2006-06-13, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing
• Universal decimal classification (UDC): 004 - Computer science and technology. Computing. Data processing

Abstract

Biološki procesi opisani su mnoštvom podataka prikupljenih promatranjima i eksperimentima. Kako bi se iz složene strukture bioloških podataka izlučile nove informacije koriste se metode dubinske analize podataka. S obzirom na složenost strukture podataka, prije same dubinske analize podatke je potrebno adekvatno pripremiti. Priprema podataka uključuje pročišćavanje podataka, integraciju i transformaciju podataka. U ovom je radu poseban naglasak stavljen na transformaciji podataka, posebno na metodama smanjivanja dimenzijske složenosti podataka. Podaci korišteni u praktičnom dijelu rada odnose se na gene za koje je potrebno predvidjeti utječu li na promatrani biološki sustav ili ne. S obzirom na problematiku, proučeni su postupci dubinske analize podataka s naglaskom na metode razvrstavanja (klasifikacije) podataka. Uspoređeni su odabrani algoritmi razvrstavanja, te je objašnjeno što je utjecalo na bolje, odnosno lošije rezultate odabranih algoritama. Uočeno je da najbolje rezultate daju algoritam VFI i algoritam 1-razred SVM.

Abstract (english)

Biological processes are described using huge amount of data collected through the observation and experiments. Data mining is used for interesting knowledge extraction. Due to data complexity, data mining process should succeed the process of cleaning data, integration of data from different sources and transformation of data. Considering the problems involved with the data set used in this work, the emphasis is set on the transformation of data, especially on the feature selection. Considering the given data set, the goal of the work is to learn a model that can predict whether the particular gene influences the specified biological system. The overview of data mining techniques is given with the special attention to classification algorithms, because the model for the given data set is build using classification algorithms. The comparison of classification algorithms showed that the best results are provided using VFI (Voting Feature Interval) algorithm and one-class SVM (Support Vector Machines) algorithm.

• Keywords: Dubinska analiza podataka; biološki podaci; smanjivanje dimenzijske složenosti podataka; algoritam razvrstavanja
• Keywords (english): Data mining; biological data; feature selection; data classification methods
• Language: croatian
• URN:NBN: urn:nbn:hr:168:969779
• Study programme: Title: Postgraduate master programe in computing; Study programme type: university; Study level: postgraduate; Academic / professional title: magistar inženjer računarstva (magistar inženjer računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2019-07-17 13:44:35