Yudin

A new approach to analysis of random processes, based on applying information theory methods and new “distribution entropy” statistics, is proposed in this paper. Use of information-theoretical methods was justified when solving several statistical problems such as tests of statistical hypothesis, analysis of process state, and so forth. It was shown that entropy is a universal statistical estimate which allows to solve almost every simulation problems and analyze complex process, including identification of the distribution law and solving discrimination problems. The main advantage of this method is a possibility of considering non-linear models with non-Gaussian factors. An example of creation and analysis of informational model of labor productivity dependence on the salary is considered.