2. SnapFast makes the fastest supercomputers in the world. The production manager has hired you as a consultant to help predict monthly demand, based on past demand figures. The following table presents the demand figures for the last 5 months. Month June July August September October Demand 30 34 3339 46 Please round to two decimal places for all computations. N-period Moving Average: F = (A-1 + A-2 + … + A-N )/N Exponential Smoothing: F = Fra + a(A-1 – F-1) Holt’s Method: FIT+1 = Ft +T Ft= QA++ (1 – a)FIT = 0A++ (1 – a)(Ft-1+T+1) T = B (Ft-Fr.) + (1-B) Tt-1 a) Based on the pattern of the above demand data, which objective forecasting method(s) would you recommend and why? b) What will be the demand forecasts for November if a 2-month simple moving average is used? c) Based on the simple exponential smoothing method with a smoothing constant, a = 0.5, calculate forecasted demand for November. Start with an initial forecast of 30 supercomputers in June d) Based on Holt’s method with smoothing constants, a = 0.5, B = 0.1, calculate forecasted demand for November. For June, start with F = 30 and T. = 4. e) Using the MAD measure, only for months from (including) August to including) October, compare the accuracy of forecasts generated by simple moving average in part (b), simple exponential smoothing in part (c), and Holt’s method in part (d). Based on your MAD measures, which forecasting method is better in this case?
EXPERT ANSWER a). Holt’s method is recommended because it provides two smoothing(level smoothing and trend smoothing). It has two smoothing constants α and β.