Adaptive targeting – analysis of forecast variants of expected reality

Abstract

Purpose: the authors describe a new method of comparative analysis for forecast variants formed using adaptive-targeted regression models and show the possibility of its practical application. Discussion: the authors provided a fairly detailed description of the adaptive regression analysis procedure and proposed a two-step procedure for comparative analysis of the autoregression model obtained after targeting with the same model before targeting based on it. The results of the analysis are usually well interpreted, allowing us to understand at a qualitative level the main differences between the past period and the expected reality, the characteristics of which are provided for in a predictive manner. The results of calculations illustrating the possibility of practical use for the proposed method in comparative analysis are presented. Results: the authors developed a procedure for economic analysis of the expected future described by the multivariate structure of the forecast image.