Open AccessMethods of linear multiple regression in a matrix form
Author(s) -
L. I Ivashnev
Publication year - 2015
Publication title -
izvestiâ mgtu "mami"
Language(s) - English
Resource type - Journals
eISSN - 2949-1428
pISSN - 2074-0530
DOI - 10.17816/2074-0530-67011
Subject(s) - mathematics , linear regression , matrix (chemical analysis) , regression analysis , regression , linear predictor function , proper linear model , total least squares , generalized least squares , design matrix , segmented regression , regression diagnostic , statistics , linear least squares , nonlinear regression , polynomial regression , linear model , materials science , estimator , composite material
The article contains a summary of three basic and two weighted linear multiple regression tech- niques in matrix form, together with the method of least squares of Gauss constitute a new tool re- gression analysis. The article contains a matrix formula that can be used to obtain equations of line- ar multiple regression and the basic weighted least-squares method to obtain regression equations without constant term and the method of obtaining the regression equations of general form. The article provides an example of use of matrix methods to obtain the coefficients of regression equa- tion of the general form, ie equation of equal performance.