Properties of Univariate Time Series
This chapter introduces a number of concepts in the analysis of univariate time series that are important for an understanding of non-stationarity in the multivariate case. The fundamental building block is the autocorrelation structure of a time series. This describes the way in which current and past values of a time series are related to one another. Capturing the main characteristics of these relationships can be thought of as the primary task of a time series model: to provide theoretical structures the properties of which closely approximate those of observed time series, and to provide estimates of such models using specific time series that can be used to draw inferences about other aspects of behaviour.
Unable to display preview. Download preview PDF.