A common way of testing forecasting procedures consists of monitoring how a particular procedure responds to a step function in demand or a one period transient change in demand level. (Needless to say, the step and transient functions are only two of the many functions that can be used.)

Time Step Function Transient Change 1 100 100 2 100 100 3 150 150 4 150 100 5 150 100 6 150 100 7 150 100 8 150 100 9 150 100 10 150 100

a. Use both a simple and a trend exponential smoothing model (trying both α = 0.1 and

α = 0.3) to track both of the above demand patterns. For initial conditions, assume at the start of time 1 (end of period 0) that aˆ = 100, ˆ

b = 0.

b. Plot the corresponding forecast errors under each procedure.