Razvrstavanje vremenskih nizova temeljeno na fragmentaciji kvalitativnog prostora

doctoral thesis

• Title (english): Time-series classification based on qualitative space fragmentation
• Author: Željko Jagnjić
• Mentor: Nikola Bogunović (mentor)
• Committee member: Slobodan Ribarić (predsjednik povjerenstva)
• Committee member: Nikola Bogunović (član povjerenstva)
• Committee member: Franjo Jović (član povjerenstva)
• Committee member: Zoran Kalafatić (član povjerenstva)
• Committee member: Ivan Petrović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Electronics, Microelectronics, Computer and Intelligent Systems); Zagreb
• Defense date and country: 2009-03-26, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing
• Universal decimal classification (UDC): 004; GENERALLY; Computer science and technology. Computing. Data processing

Abstract

U ovom radu prikazana je nova metoda razvrstavanja vremenskih nizova koja se temelji na fragmentaciji kvalitativnog prostora. Osnovu metode čine postupci potpunog oduzimanja i kvalitativnog kodiranja. Takvim se pristupom kvantitativne veličine vremenskih nizova preslikavaju u kvalitativan prostor koji se predočuje kvalitativnim matricama. Kvalitativne matrice omogućile su nov način predstavljanja i analize vremenskih nizova koji je blizak ljudskom poimanju podataka, a temelji se na vizualizaciji podataka i opisivanju slikovnih oblika. Predloženi osnovni skup simboličkih i numeričkih značajki iskorišten je u procesu strojnog učenja, odnosno kreiranju stabla odlučivanja te procesu razvrstavanja. Testiranjem na različitim vremenskim nizovima pokazane su prednosti i nedostatci nove metode u odnosu na poznate postupke.

Abstract (english)

In this work a new time-series classification method based on qualitative space fragmentation has been presented. The main features of the proposed method are global differencing and qualitative coding. With such an approach, quantitative data of the considered time-series are translated into qualitative space, which can be observed through qualitative matrices. Qualitative matrix form provides the means for a new approach to time-series analysis, which is closely related to human way of data interpretation based on data visualization and shape description. Basic set of symbolic and numeric data features have been defined and used in machine learning process through the decision tree induction and classification. Extensive testing with different time-series has revealed key advantages of the presented method, but also exposed some limitations as compared to other approaches.

• Keywords: potpuno oduzimanje; kvalitativan prostor; fragmentacija; vremenski nizovi; izdvajanje značajki; stabla odlučivanja; razvrstavanje
• Keywords (english): global differencing; qualitative space; fragmentation; time series; feature selection; decision trees; classification
• Language: croatian
• URN:NBN: urn:nbn:hr:168:724741
• Project: Number: 036-0362980-1921; Title: Računalne okoline za sveprisutne raspodijeljene sustave; Leader: Siniša Srbljić; Jurisdiction: Croatia; Funder: MZOS; Funding stream: ZP
• Study programme: Title: Computer Science; Study programme type: university; Study level: postgraduate; Academic / professional title: Doktor znanosti (Doktor znanosti)
• Catalog URL: http://lib.fer.hr/cgi-bin/koha/opac-detail.pl?biblionumber=33342
• Type of resource: Text
• Extent: 149 str.
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-03-24 11:56:29