The conceptual basis of obtaining and study maximum time of the forecast with a predetermined accuracy

Abstract

Purpose: the author constructs systematically summarizing combination of cluster analysis models and Kendall concordance models for aggregation the values of different private forecasts. Discussion: the author introduces the term of «predictability» for the different nature time series. Note that the term of «predictability» first justifies the imperative. This imperative links the length of the forecast horizon, the specified forecast quality and statistical and fractal characteristics of a predictable process. Mathematical formulation of the problem, as well as models, methodologies and assessment ways the degree of «predictability», instrumentation and software must have algorithms and tools for its implement. These algorithms and tools are the part of professional systems of computer mathematics. Results: the author related the problems of quantitative evaluation of «predictability» directly with the «horizon of the future» length. The writer obtained a priori estimates of longitude and forecast quality depending on the statistical properties of the predicted signal at preassigned precision.