The basis behind the AR model comes from the idea that the output/dependent variable is
a linear function of its previous values as lag variables. The easiest way to understand this
is via example: the simplest case of an AR model is AR(1): Yt = c + 1Yt