Building ARIMA time series models
The term ARIMA is made up of the letters that represent a modeling approach for time series data. ARIMA models contain the following three elements:
AR: Auto regressive, specified with p or P
I: Integrated (differencing), specified with d or D
MA: Moving average, specified with q or Q
Auto regressive means that earlier lagged points in the data influence later points in the sequence. This creates a dependence condition. The type of AR model chosen is based on how many steps away (lags) the points in the past affect the points in the future. Data that has a greater lingering effect on future points has a higher lag. The higher the lag, the higher the AR number. You will see models referred to as AR(1), AR(2), and so forth to represent an autoregressive model of the number of p lags specified in the parentheses.
Integrated refers to differencing that you learned earlier. The d value represents the number of differences used in the model. It is typically...
