n the class, we assumed that the demand process is stationary of the form Dt = t where 's are independent and have mean 0 and variance 2. Instead, suppose that the demand process has a linear trend Dt = a b.t t for some unknown parameters a and b (t is time period); however, the moving average forecasts and the exponential smoothing forecast are computed the same way. We want to analyze the performance of the 2 models under this new "world" where demand is not of the form we assumed in class