Homobile COmpany Case
...Forecast: Quarter 26: 796.6801 orders/quarter CFE = -175.5640 Quarter 27: 898.4452 orders/quarter MAD = 34.8544 Quarter 28: 870.4586 orders/quarter MSE = 1957.8630 Quarter 29: 934.2638 orders/quarter MAPE = 5.9212 Trk Signal = -5.0371 R-Square = 0.8349 Standard Error = 1.25 (34.8544) = 43.568 Degree of Confidence: Quarter 26- 68% sure orders between 753 and 840 95% sure orders between 710 and 884 99% sure orders between 666 and 927 Quarter 27- 68% sure orders between 855 and 942 95% sure orders between 811 and 986 99% sure orders between 768 and 1029 Quarter 28- 68% sure orders between 827 and 914 95% sure orders between 783 and 958 99% sure orders between 740 and 1001 Quarter 29- 68% sure orders between 891 and 978 95% sure orders between 847 and 1021 99% sure orders between 804 and 1065 E) Our recommendations for Homobile would be to use the HWA method to forecast while allowing WinQSB to determine the best smoothing coefficients. Also, during seasonal periods when on time delivery is most important to the consumer and opportunity cots are higher, management should error on the side of over predicting errors. When on time delivery is not as important to the customer and opportunity costs are reduced, as in the non-seasonal period, management should error on the side of under predicting. We also recommend that Homobile continuously updates their actual demand for quarterly orders and that they use updated forecasts for decision making in the future, even for the quarters already forecasted. Management should also look into what factors the life cycle plays on demand and what they can expect entering into the next phase. In addition, competitive advantages such as pricing, quality, and on time delivery sensitivity in the marketplace should be assessed to predict how costs will effected in the future. F) The method of forecasting is the concrete manner in which historical data calculated using various methods such as SES, SEST, HWM, and HWA to produce numerical predictions for future demands. The methodology of forecasting on the other hand is not as concrete as the method because forecasting is more of an art than a science because what a forecast is trying to do is predict future outcomes and there is no one single concrete method for doing that. When forecasting the decision maker not only uses their expertise in statistics and mathematics to come up with a forecast, but they often also use emotions, intuitions, and personal experience with the situation to aid in their forecasts. The role of the computer in forecasting is to assist with the statistical/ mathematical calculations like WinQSB did for us. The computer simplifies the process of forecasting and analyzing by increasing the speed and accuracy at which we can compute the information that is valuable for our forecasts (i.e. plotting graphs to show seasonality and trends). The role of analysts in the forecasting system at a modern corporation is to constantly be collecting and analyzing data that is pertinent to their company’s performance such as factors in the industry, economical conditions, market conditions, and any other variable that could possibly have an effect on future outcomes for the company. These analyses are then used to create “educated” forecasts through programs such as Minitab, SPSS, and WinQSB which make predictions based on all of the pertinent information that was available to the analyst. The analyst then has to look at those forecasts using their intuition, emotions, and expertise and determine how accurate they think they are and how much weight should be assessed to them. CompanyHomobile Company Case A) During the 2nd Quarter of 1999 the normal sales level was disrupted by “the blizzard of the century” and 190 additional trailer homes were ordered by an insurance company to settle with their customer’s claims of ruined trailer homes from the intense storm. Since storms of this size rarely occur it would be best to adjust the unusual data point to a sales number that would be closer to the norm for that quarter so as not to throw off forecasts with one huge random outlier. By looking at Graph 1 we can see the upward trend and seasonality of the trailer sales over time. To bring the unusual data point to a more reasonable value, the 190 additional sales from the blizzard were subtracted and then a growth rate of 2.5% was added for that quarter. Therefore we adjusted the unusual data point from 730 down to 565 to better fit a normal growth for that period. Note that the growth rate prior to the adjustment for that quarter was 32.5% (Appendix #1). B) By once again looking at Graph 1, we can see that both seasonality and trend exist for Homobile’s sales of trailer homes. In the long run demand is generally increasing although there are fluctuations when you focus on smaller increments of time. This long-term upward progression is our trend. Seasonality exists because those short term fluctuations in sales repeat. In looking at Graph 1 more closely we can see that each fiscal quarter is similar in nature to next year’s corresponding quarter. (i.e. Quarter 9 (II 2002) and Quarter 13 (II 2003) in Graph 1). The seasonality continues throughout the historical data but a shift begins to occur in which quarter has the highest number of orders. This may be due to an accelerated product life cycle of the trailer homes. The patterns that arise from the historical data do seem to be counter-intuitive because the periods that have the highest demands (Quarters 1 and 4) occur from October to March which is the off-peak season for traveling and vacationing. Our intuitions would lead us to believe that demand would be higher in the summer months when vacationers are most active but this is not so. One possibility for the increase in demand during off-peak vacationing time is that it takes about three months from the time you order the trailer to when you receive it. In this case you would be lead to believe that Homobile customers are simply thinking about their vacationing plans in advance and want to have their trailers ready for those peak vacationing months of the year. C) We decided to use the MAD error metric because of the way Homobile’s costs will be structured. We believe that there will be linear costs of under and over prediction for the company. The costs of under predicting are mainly due to the per unit contribution margin lost from a missed sale. The costs associated with over prediction are mainly inventory related. For example, the increasing cost of insurance and storage space on a per unit basis. Note that both sides of over and under predicting are linear but costs with some confidence will most likely not be equal between over and under predicting. For example, during high seasonal periods, we can expect that the opportunity cost of under predicting will be greater during non-seasonal periods. If Homobile offers their customers superior price advantage or quality compared to its competitors, consumers during the non-seasonal period may be willing to wait another three months for their product due to the fact that the next peak vacation period is greater than three months away. Thus, during non-seasonal periods opportunity cost is reduced. In this case it may be cost beneficial to want to over predict in the seasonal period and under predict in the non-seasonal period. D) Forecast: Quarter 26: 796.6...