Full Article: PDF
Scientific Object Identifier: http://s-o-i.org/1.1/TAS-02-106-64
DOI: https://dx.doi.org/10.15863/TAS.2022.02.106.64
Language: Russian
Citation: Zhanatauov, S. U. (2022). Method for regulating the proportion of dominant eigenvalues for a fixed matrix of eigenvectors. ISJ Theoretical & Applied Science, 02 (106), 601-613. Soi: http://s-o-i.org/1.1/TAS-02-106-64 Doi: https://dx.doi.org/10.15863/TAS.2022.02.106.64 |
Pages: 601-613
Published: 28.02.2022
Abstract: In the article, quantitative parameters and variables of the method for regulating the share of dominant eigenvalues are found for a fixed matrix of eigenvectors of the correlation matrix Rnn. Changing the variances (eigenvalues from the matrix ?nn) and increasing the proportion of variances f4(?nn,2)=(?1+?2)/n did not lead to an increase in the number ?=2 of dominant dispersions (according to the Kaiser-Dickman criterion), but increased the share of extracted knowledge : an increase in the values of off-diagonal elements by 20% leads to parallel shifts of the curves (y-variables) and to an increase in the share of information by 7.4%, from which hidden knowledge is extracted. The hidden knowledge extracted from the triplet of matrices (Сnn,R(s)nn,?(s)nn) refers to the calculated matrices of z-variability values Zmn={zij} of correlated z-variables) and to the matrix Ymn=ZmnСnn={yij} y -variability of uncorrelated y- variables.
Key words: matrix, variability, eigenvalues, eigenvectors.
|
