I.E. Tarasov

The parameters of the processes running in complex systems can represent the results of indirect measurements or the parameters of retrogression dependences of directly measured values. The widespread method of retrogression de-pendences is the least-squares method (LSM) which application is restricted by linear (or linearized) processes as the simplest for analysis. The method suggested in the article based on their probabilistic estimation according to Bayes theorem doesn-t suppose direct analytical expressions for unknown value determination demanding instead a numerical analysis computationally intensive, it is however used for free type retrogression dependences. The additional advantage is the possibility of taking into account the measured quantity distribution law.