Review article

The article presents a method for forecasting the parameters of building structures using regression equations. The most commonly used regression equations were used, namely, linear, polynomial, power, exponential, exponential, logarithmic, semi-logarithmic, hyperbolic and logistic ones. The authors propose to use one-factor regression equations in which the variable value is time, and the dependent value is the parameter of the building structure, the changes of which the researcher needs to determine. Thereafter, the authors present the basic equations as multi-factorial ones. This is achieved by means of replacing the coefficients of the main equation with regression equations which are obtained after carrying out a series of tests with variable values of the selected input parameters (such as environmental conditions, material of construction, etc.). Regression equations are derived for each state, and, as a result, a number of parameter values for the basic regression equations are available. Then, a repeated regression analysis is carried out and a regression equation is set up for the coefficients of the main regression equation, which depends on the value of the specified parameters. Such equations are called secondary equations. Examples are given for conditional linear regression, where it is demonstrated how the coefficients of the main regression equation are replaced and what final form the equation achieves after replacing the coefficients with secondary equations. The presented method allows, in the case of expression from the basic equation of the time parameter, to forecast the residual resource, with a large number of parameters of building structures determined.