# Industrial Smoothing Case Study

874 Words4 Pages
Comparing the error estimates among four forecast methods (total demand): Forecasting method MAD MAPEt TS Range Four period Moving average 13,442 13.19 -10.82 to -1.00 Simple exponential smoothing 16,403 19.32 -5.99 to 6.16 Holt’s model 6,475 7.52 -4.01 to 3.84 Winter’s model 5,227 5.56 -4.62 to 5.99 Table 18: Error estimates for HSG steel sheet total demand forecasting It’s obvious from the table 17 that the two simplest methods are useless for the total demand. This shows that the demand includes a observable growing trend, which make the Moving average method always underforecasts the real demand. Holt’s model and Winter’s model in this case both produce the good result but the Winter’s model proves its superiority due to the lower error.…show more content…
The Trend and seasonality-corrected exponential smoothing method (winter’s model) proves its best accuracy in the four methods with the lowest MAD and MAPE. The two partial demands Domestic market demand and Export market demand show the different seasonality. And the Winter’s model is appropriate for both of them. However, the Winter’s model is even more accurate when applying for the total demand which combines these two different seasonal demands. The MAPE 5.56% produced by the Winter’s model in this paper is better than the MAPE 8% produced in “Supply Chain Management” written by Sunil Chopra and Peter Meindl for the Tahoe Salt case. Therefore, in this steel sheet demand case, the company should forecast the total demand instead of forecasting the two market demands separately. 4. Adjusting forecast Winter’s model is chosen to forecast the future demand for the total market with the following forecast results: F17 = 109,494 tons F18 = 139,773 tons F19 = 131,258 tons F20 = 140,829 tons With MAD = 5,227 and MAPE =…show more content…
Which is in the same point with the safety rules of Neal Wagner, Zbigniew Michalewicz, Sven Schellenberg, Constantin Chiriac and Arvind Mohais (2011) (5). With the up-trend demand data like HSG’s steel sheet demand, the safety rule is given by: “If an up-trend is detected, the ﬁrst three weekly forecasts must be at least as high as the most recent historical demand seen.” Because there is a seasonal factor in this data so the rule will be: every quarter needs to be higher than that of last year. And the forecasts already follow the