Testování efektivnosti nových grafických metod pro identifikaci odlehlých pozorování

Abstract

This diploma thesis is focused on the practical application of new graphical methods for outlier identification. It includes a comparison of classical and new boxplot methods on datasets generated for different probability distribitions and testing probability distribution plots for selected non normal probability distributions. A substantial part of the diploma thesis is solved as SGS project "Testing the effectiveness of new graphical outlier detection methods".