Machine Learning With Go

Auto-regressive models for forecasting

The first category of models that we are going to use to try and forecast our time series are called auto-regressive (AR) models. As already mentioned, we try to model a data point in our time series based on one or more previous points in the series. We are, thus, modeling the time series using the time series itself. This use of the series itself is what distinguishes AR methods from the more general regression methods discussed in Chapter 4, Regression.