A new version of the least squares method for multifactor linear regression analysis and its applications

The new variant of the least square method for linear regression analysis is presented. The idea of the new version is minimization of the sum of squared distances between the hyperplanes and the given statistical points. The comparing analysis of new method and classical algorithm for dispersion is presented. The non-uniqueness of the solution of the problem of choosing a linear multifactorial function is shown. Numerical examples illustrating the main ideas of the new approach are given.