Predictive Analytics: Microsoft® Excel

9. Box-Jenkins ARIMA Models

In This Chapter:

The Rationale for ARIMA

Stages in ARIMA Analysis

The Identification Stage

The Estimation Stage

The Diagnostic and Forecasting Stages

Suppose you’re fortunate enough to have at hand a nice long baseline of observations. You’d like to model how the data behaves over time, and if possible create a credible forecast. You have your choice of approaches for that task. As Chapter 4, “Forecasting a Time Series: Smoothing,” and Chapter 5, “Forecasting a Time Series: Regression,” showed, you could take a smoothing approach—that is, a variation on moving averages—or a regression approach, possibly one based on autoregression so that you’re using the baseline to forecast itself.

How do you choose which approach, ...