ARIMA modeling in practice
As with any theoretical framework, there are subtleties and nuances to getting ARIMA to work in practice. We cannot cover all the tips and tricks of ARIMA modeling in practice in this short section. Instead, we will focus on the practical issues related to what we introduced in the previous section, namely, how to identify the order 
that we should use when building an ARIMA model of a given time series dataset.
Unit root testing
The first stage in deciding the appropriate orders of an ARIMA model is to determine whether the time series data you have represents an integrated series or not. This is usually done by testing for the presence of what is called a unit root. We won’t go into the full details of what a unit root is, other than to say it is a root of a polynomial equation. A unit root has magnitude 1 and the polynomial equation is constructed from coefficients of an AR(p) model estimated from the time series. The presence of a unit root...
