Tutorial paper V: Analyse des sertes chronologiques: approches temporelles et fréquentielles

Abstract

The aim of this paper is to present the basic principles of time series analysis. We first introduce the concepts of linear stochastic models and spectral functions associated to a stationary process. In a second part, we consider the statistical analysis of time series in the stationary case. In a third part, we study the problems resulting from the assumption of stationarity and we consider generalizations of the previously presented methods. In this paper, we consider only univariate analysis. Other problems (namely forecasting) will be presented in another paper.