Time Series and Forecasting
D.H. Stamatis · 2002
Forecasting is very difficult but essential in any business. Most forecasts are based on past history, and there lies one of the problems. There is no guarantee that the past will repeat itself. Another, of course, is the introduction of changes that will certainly change the outcome of the expectations. Practically all forecasts get these two principles wrong and, as a consequence, statisticians depend on the confidence levels of the forecast to save face. This chapter will introduce the reader to time series and forecasting in a cursory way. Specifically, it will cover autocorrelation, exponential trends, curve smoothing, and econometric models.