• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 1
  • Tagged with
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

Statistinė medicininių duomenų analizė taikant dimensijos mažinimo metodus / Statistical analysis of medical data using methods of dimension reduction

Šlepikaitė, Laura 25 November 2010 (has links)
Darbe yra aprašomi skirtingi dimensijos mažinimo metodai. Pradžioje pristatoma pagrindinių komponenčių analizė ir ypatingųjų reikšmių dekompozicija. Po to plačiau nagrinėjama neneigiamos matricos faktorizacijos tema, aprašomi algoritmai naudojantys atnaujinimo dauginant metodus. Vėliau pasiūlomi algoritmai naujai prie duomenų matricos prijungiamų duomenų transformacijai atlikti. Visi algoritmai realizuoti naudojantis SAS statistiniu paketu. Tyrimui naudoti pacientų, sergančių širdies nepakankamumu, duomenys. Gauti rezultatai parodo, kad dimensijos mažinimas gali būti efektyvus įrankis dirbant su didelio matavimo duomenų rinkiniais. / This work deals with several dimensionality reduction techniques and their implementations in real medical problems. For this reason, firstly, one speaks about classical dimension reduction methods called principal component analysis and singular value decomposition. After these methods are introduced, non – negative matrix factorization (NMF) are presented. Also algorithms for its implementation are introduced. Moreover, two ways for implementation of dimensionality reduction via NMF are presented when applied for feature extraction, followed by pattern recognition. All algorithms were executed using SAS statistical pachage. Patients with heart failure data were used. It was shown that dimensionality reduction could be effective tool for multidimensional data analysis and classification problems.

Page generated in 0.0768 seconds