When should an analyst use an ARIMA model versus the ARMA model for stationary datasets?
Here’s what I wrote, did I miss anything?
Both the ARIMA (Auto-Regressive Integrated Moving Average) and ARMA (Auto-Regressive Moving Average) models can be used to make forecasts on stationary data; however, the characteristics of the dataset determine which model is ideal.
If the dataset is stationary and exhibits a clear trend or seasonality, the ARIMA model is a better choice, as the model already has differencing to remove the trend/seasonality components. The “integration” component of ARIMA will be able to capture any non-stationarity in the residuals. Therefore, any model that needs differencing applied to the data would be better suited with the ARIMA model.
One may use the ARMA model if the dataset is stationary and does not need differencing to achieve stationarity. The ARMA model is simpler than the AIRMA, so if differencing is not needed, ARMA is preferred.