The appropriateness of stationary autoregressive processes model for the calculation of "exact" values of the adaptive lattice filter (ALF) parameters using is justified. The approximations of correlation ALF partial coefficients distributions are obtained. The dependences of these parameters on the volume of training sample standard deviation estimates and the displacement of their expectation are found. The necessity of termination adaptation (parameter estimation) in the "senior" levels of ALF is statistically justified.
statistical analysis, adaptive lattice filters, partial correlation coefficient, the process of autoregression, correlation sequence