Využití parametrické a neparametrické regrese při identifikaci odlehlých hodnot a nevalidních měření

0447951 - ÚVGZ 2016 eng C - Conference Paper (international conference)
Mikušková, M. - Dvorská, Alice - Michálek, J.
Využití parametrické a neparametrické regrese při identifikaci odlehlých hodnot a nevalidních měření.
[The application of parametric and nonparametric regression for identification of outliers and invalid measurements.]
[Mezinárodní kolokvium o řízení vzdělávacího procesu /13./. Brno (CZ), 28.05.2015-]
R&D Projects: GA MŠMT(CZ) LO1415
Institutional support: RVO:67179843
Keywords : linear regression * robust regression * kernel regression * concentrations of tropospheric ozone in the air
Subject RIV: DG - Athmosphere Sciences, Meteorology

Relatively large differences occur sometimes between time proximal measurements during the continuous detection of some pollutants. The acceptability of these differences is identified by using appropriate statisticalmethods and subsequent subjective judgment of person specialized in appropriate measurement. This article presents the results of linear, robust and kernel regression that were applied on measured concentrations of tropospheric ozone in order to identify outlier values and flag invalid measurements.
Permanent Link: http://hdl.handle.net/11104/0249683