Predictive Analytics: Microsoft® Excel

5. Forecasting a Time Series: Regression

In This Chapter:

Forecasting with Regression

Forecasting with Autoregression

Chapter 4, “Forecasting a Time Series: Smoothing,” discussed one general approach to forecasting: exponential smoothing. Whether you use simple exponential smoothing, Holt’s method or another approach to exponential smoothing, each actual observation has some weight in the value of the next forecast. Older observations carry less weight, and new observations more weight, because the older the observation, the larger the exponent that’s applied to the smoothing fraction. For example, 0.2 raised to the 30th power puts much less weight on a value that’s thirty periods old than does 0.2 raised to the 5th power on a value that’s ...