Open AccessCLASSIFICATION OF MULTIVARIATE SAMPLES USING PETUNIN ELLIPSES
Author(s) -
D. A. Klyushin,
Ya. V. Shtyk
Publication year - 2020
Publication title -
žurnal občislûvalʹnoï ta prikladnoï matematiki
Language(s) - English
Resource type - Journals
eISSN - 2706-9699
pISSN - 2706-9680
DOI - 10.17721/2706-9699.2020.1.05
Subject(s) - ellipse , multivariate statistics , variance (accounting) , quadratic equation , mathematics , statistics , pattern recognition (psychology) , computer science , artificial intelligence , geometry , accounting , business
The method of classification multivariate samples using Petunin ellipses is investigated in the paper. Several different types of samples were generated for testing. Based on the calculated accuracy of the criteria advantages and disadvantages of each of the linear and quadratic criteria and the specifics of the method as a whole were discovered. It has been found that both linear and quadratic criteria give high accuracy for samples with small variance. As the variance increases, the accuracy of the linear criterion remains high, the accuracy of the quadratic criterion decreases. Both criteria are resistant to sample noise.