Linear Models

Abstract

consistency changes of several variables on the assumption that their empirical values are not identical to "true". Such situations arise, for example, because of the presence of errors registration of empirical data, due to the measuring equipment. As a typical example, you can specify the task of identification when determining the coefficients of the linear model of input -output. You may also notice a situation when valuation parameters obtained during the training phase, are used in the processing of data recorded in different time intervals, for example in the problems of pattern recognition or self-compensation interference. In some cases we can speak about models of the interaction (relationship) of the studied processes. Explore the feasibility of using in these conditions the principle of orthogonal projecting of empirical values on the hyperplane defined by a set of analyzed the "true" values. Received basic computational formulas for estimating the parameters of linear models of consistency.