On Applicability of IQR Method for Filtering of Experimental Data
Β. B. Ilyushin
Abstract
Abstract The results of testing the popular IQR (Interquartile Range) method for filtering experimental data are presented. It is shown that if the distributions of measured values differ greatly from the Gaussian distribution, this method gives a large error in the statistical characteristics, especially the higher moments. The earlier-developed statistical filtering method can take into account substantial skewness of distributions of measured values and can greatly reduce the filtering error.
Topics & Concepts
Computer scienceRange (aeronautics)Materials scienceStatistical physicsPhysicsComposite materialFault Detection and Control SystemsAnomaly Detection Techniques and ApplicationsAdvanced Statistical Methods and Models