Time Series Modelling in R
Alfred Kume · 2023
Time series data are generally considered as a set of observations ordered in time, typically at equally spaced intervals such as days, hours, minutes, and months. The main aim of the time series analysis is to understand the driving dynamics of the data generating process and use this for forecasting future values. The main difference between linear regression and basic time series models is based on the way the dependence among observations is addressed. Intuitively, stationarity is present in the time series data if the key statistical properties – the mean, variance, and pairwise dependence of data points – are invariant of the time window of observations. The first step in modelling stationary time series data is to explore the autocorrelation structure since it provides further insight into the underlying time dependence. As indicated in the standard theory, the method of moments or the Yule-Walker estimator, is applicable only for the pure Auto-regressive models.