Exponential smoothing of time series data in R

This thesis deals with basic methods of classical exponential smoothing for univariate time series with regular observations and their applications in an open source statistical computing environment of a programming language R. The methods of simple, double and triple exponential smoothing, Holt linear exponential smoothing and Holt-Winters exponential smoothing are presented. The recursive character of the methods along with their high computational effectiveness are demonstrated. The latter part of the thesis demonstrates application of the exponential methods in the R language environment, based on selected time series data from the Time Series Data Library (TSDL). Short term forecasts of the series are constructed and ilustrated graphically using selected functions of the R environment.