MATH MODELING

Statistics construction algorithms for non-stationary time series analysis and forecasting

K. P. Osminin

Lomonosov Moscow State University, Faculty of Mechanics and Mathematics

Abstract: The numerical algorithms for construction of the horizontal statistics for any given non-stationary time series are described. The maximal interval of forecasting for given accuracy of empirical density of distribution function is determined. The algorithm of evolution of empirical density is realized for the prognosis of non-stationary quasi-distribution. The example of forecasting of financial instruments is discussed.

Keywords: non-stationary time series, horizontal statistics, optimal set, forecasting, kinetic equations.

Bibliographic databases:

•