Total Msplit estimation

Abstract M split estimation is a method that enables the of mutually competing versions parameters in functional observation models. In presented study, classical models found it are replaced by errors-in-variables (EIV) Similar to weighted total least-squares (WTLS) method, random components these were assigned covariance matrix Thus, proposed named Total (TM ) estimation, corresponds basic rules WTLS. TM objective function constructed using squared and WTLS functions. The algorithm based on Gauss–Newton applied linear approximation EIV properties examples regression line two-dimensional affine transformation.