University of PardubiceFaculty of Economics and Administration

University

Employees
|
Students
Česky

Publication detail

Analýza informačního systému pomocí rough množin

Authors: Jirava Pavel

Year: 2007

Type of publication: ostatní - dizertace

Publisher name: Univerzita Pardubice

Place: Pardubice

Page from-to: nestránkováno

Titles:

Language	Name	Abstract	Keywords
cze	Analýza informačního systému pomocí rough množin	Předložená disertační práce pojednává o možnostech využití teorie rough množin pro analýzu neurčitých dat v informačním systému, který je reprezentován informační nebo rozhodovací tabulkou. Zabývá se analýzou informačních systémů, syntetizuje a analyzuje navrhnutý hybridní rough-fuzzy model klasifikátoru s využitím realizovaného softwarového nástroje Rough Fuzzy Toolbox.	informační systém;rozhodovací tabulka;teorie rough množin;teorie fuzzy množin;funkce příslušnosti;data mining;klasifikace;generování pravidel;rough-fuzzy hybridní model;neurčitost.
eng	Information System Analysis based on Rough Sets	The submitted thesis focuses on the models proposals and rough sets theory application for information system analysis, and the contiguous possibilities of the uncertain and missing data processing in them. It seems to be effective and up-to-date to tackle this problem using the theory of rough sets and a hybrid approach combining rough sets and fuzzy sets, both belonging to the field of the computational intelligence research. The rough sets theory which is the backbone of this thesis is based on the research of information system logical properties, and uncertainty in it is expressed by a boundary region. Every investigated object is connected to a specific piece of information, to specific data. The objects which are characterized by the same pieces of information are mutually undistinguishable from the point of view of the accessible pieces of information. This is expressed in the rough sets theory by the indistinguishableness relations. The thesis summarizes and explicates the status quo of the investigated problem and the methods suitable for information system analysis. Further, it deals with up-to-date methods for uncertain data operating, for information system evaluation, for rules generating and the consequent classification. The proposed algorithms and hybrid rough-fuzzy classifier model are carried out in MATLAB, and tested on more data files, and compared to others, already known methods.