Moving average model

A moving average model (MA) is a linear combination of historic white noise error terms. Let's have a look at the equation of the model:

Here, ω is the white noise with E(ω_t)=0 and variance = σ². 

In order to find out the order of the AR model, we need to plot a partial autocorrelation function plot, and then look for the lag where the upper confidence level has been crossed for the first time.