Non-stationary time series: Forecasting methods with examples of analysis of financial and commodity markets

This book describes the methods of analysis and forecasting of time series, which are found in practical activities. Both traditional methods developed for stationary time series and new approaches are presented, which are proposed to be used to analyze non -stationary random processes, if the use of standard adaptive procedures to the latter does not provide the necessary forecast accuracy. The main goal of the book is to propose to practicing analysts an instrument of studying time series, based on some empirical statistics and has certain theoretical justifications. The developed approach is based on the concept of a selective function of distribution that changes over time.
The book is conditionally divided into two parts: theoretical, which sets out the necessary information from the theory of stationary and non -stationary random processes and mathematical statistics, and practical, where examples of the application of the developed theory for the analysis and forecasting of time series found in various fields of activity are given