Pages: 501-514
Published: 30.10.2018
Abstract: The article describes the stages of calculating the model values of the matrix elements Zmn = [Z1?Z2] of centered and normalized values n = 4 z-variables, which have assigned values of the regression coefficients, ?1,?2,…, ?n-1 exactly equal to the calculated by real matrix values. The partition of the matrix [Z1?Z2] corresponds to a regression model of the form zn=?1z1+?2z2+…+?n-1zn-1, where z1,z2,…,zn-1 is a set of explanatory (independent) variables ("regressors"), zn - response variable (dependent variable), ?1,…,?n-1-regression coefficients. The values of 4 valid performance indicators of 17 branches of the bank Z17,4=[Z1|Z2] are considered. The MLRA-samples Z(t,?)17,4=[Z(t,?)1|Z(t,?)2] are modeled for assigned values of the regression coefficients. The features of the application of IM MLRA [1] at n=4 are found. Model MLRA samples (for m=17,n=4) Z(t,?,)mn=[Z(t,?,)1?Z(t,?)2] satisfy both DM MLRA and IM ILRA:(1\m)Z(t)Т1Z(t)1=R(?)11, (1\m)Z(t)Т1Z(?,t)2= R(?)12, R(?)12=R(?)11?, ?=(?1, ?2,?3)T, adequacy to the future, not the past, real multidimensional sample Х017,4.
Key words: assigned values of regression coefficients, model of multidimensional linear regression analysis of valid indicators.
|
