Assignment 1
Question 1: The customer order decoupling point is described as the point at which demand changes from independent to dependant (unit 2). What does this mean and why is this important to managers? This means that the customer order decoupling point (CODP) is the point at time which the company has to allocate resources(such as timing and quantity of material) in order to serve the final consumer(the independent demand that now becomes dependent) requests. Different environments are defined according to different CODP, they are: * The Make-to-stock (MTS) environment has its CODP in the finished goods inventory, it means that the buyer will buy the product directly from the inventory. The focus of the demand management is providing finished goods when and where the customers want them. In order to have assure these customer service (control) said before, the demand management have to plan inventory levels that will supply the demand correctly(planning) using and providing forecast (information). * In the Assemble-to-order (ATO) environment, the CODP is when the consumer decide what already produced components he wants and the company would then assemble them to complete the consumer's order. In order to meet delivery dates (control), the demand management has to determinate these delivery dates (planning). The demand management also has to know if the components can (or can't) be combined in a viable product, it means: see if the combination asked by the client is viable, it is called configuration management(information). * The CODP of a Make-to-Order (MTO) environment is the raw material inventory fabric. The focus of the demand management is coordinating information on customers' product needs with engineering. In the Engineering-to-Order (ETO) the CODP is with the suppliers. The difference between these two environments are the costumer order decoupling point and also because in the ETO environment more of the detail design may be left to the engineers than the customer. Forecasting to determine engineering capacity is fundamental to demand management. The information is the product specifications. Provide engineering capacity (planning) and adjust capacity to customer needs (control) are also tasks of demand management. To illustrate it: One costumer go and buy a refrigerator from the inventory of store, it's a MTS environment and the CODP is in the finished refrigerator. If the consumer calls to Apple and asks for a computer with 8 GB memory, 1 TB HD, an i7 processor, it's a ATO environment and the CODP is the work-in-process parts and components. If the consumer goes to a restaurant and ask for a meal with meat, rice and beans, it's a MTO environment and the CODP is the raw materials. If a consumer decides to build a house and goes to BCI Australia, it's a ETO environment and the CODP is the suppliers that will provide BCI the materials to build the house. Furthermore, the CODP is fundamental to demand management because different environments request different forms of information, planning and control to attempt the demand, and a key concept in these environments is the costumer order decoupling point. The manager, in order to supply the requirements of this new dependent demand (the final consumer), has to plan and control according to the CODP. Question 2: ME, MAPD, Cumulative Error – what do they tell us and why do we want to know? Forecasting is a really important area in the demand management. Although the forecasts are always wrong, it is an important tool used to guide the demand management because it can provide more accurate information about the demand in the future. In order to evaluate the forecasting and see if it is accurate enough, we have to compare this forecasting to de real data. That is where the errors have a important role. For any forecasting procedure, an important criterion is honesty, or lack of bias. The mean error (ME) is used to measure bias. In ME, forecasts that were too high are offset by forecasts that were too low, i.e, a good forecasting would have to be as close to zero as possible. So, the ME tell us if the forecast in determined period was accurate enough to the reality in that period. However, sometimes it is not enough, for example: Month | 1 | 2 | 3 | 4 | Demand | 10 | 11 | 15 | 10 | Forecasting | 7 | 7 | 10 | 22 | Although the ME is zero, the discrepancy each month is huge and the company did not sell as much as asked by the market and now have a full inventory. So, demand management is concerned about the error magnitude. In order to evaluate the accuracy of the forecast is used the mean absolute percentage deviation (MAPD). Different from the ME, it considers the error magnitude. The MAPD tell us how accurate was our forecasting in period of time. A good forecasting has the MAPD as close to zero as possible. Actually, a forecasting with a MAPD = 0 is a "perfect" forecasting because all the demand was exactly equal to the forecasting. For example: Month | 1 | 2 | 3 | 4 | Demand | 10 | 11 | 15 | 10 | Forecasting | 7 | 7 | 10 | 22 | The MAPD is 55%, it show us that our forecast was really not accurate and the company would lose money if it was used. Although MAPD is a really good tool to measure the accuracy of a forecast, it has problems that big or small errors in one forecasting might distort the MAPD. Another tool used to evaluate forecasting is the Cumulative Error. It show us how wrong was our forecasting in a period, not considering the magnitudes or . The difference between the Cumulative Error and the Mean Error is that the Mean Error is an average of errors between forecasting and reality in a period and the Cumulative Error considers only the sum of errors in that same period. Consequently, the Cumulative Error is always higher or equal to ME. Furthermore, they are all tools to measure how close to reality is our forecast. The ME measures bias and is an average of the errors in a period, the cumulative error is a sum of errors in a period and the MAPD is a average of discrepancies in a period. We want to (actually, have to) know it in order to choose the best forecast to produce and supply our demand correctly.

Question 3 a) Use a three-month moving average to estimate the month-in-advance forecast of demand for months 4-12 and generate a forecast for the first month of next year. Calculate the Mean Error and the Mean Absolute Error. Month | Demand | MV3 | MV3(arredondado) | ME | MAE | 1 | 20 | - | - | - | - | 2 | 18 | - | - | - | - | 3 | 21 | - | - | - | - | 4 | 25 | 19.67 | 20 | 5 | 5 | 5 | 24 | 21.33 | 22 | 3 | 3 | 6 | 27 | 23.33 | 24 | 4 | 4 | 7 | 22 | 25.33 | 26 | -3 | 3 | 8 | 30 | 24.33 | 25 | 6 | 6 | 9 | 23 | 26.33 | 27 | -3 | 3 | 10 | 20 | 25.00 | 25 | -5 | 5 | 11 | 29 | 24.33 | 25 | 5 | 5 | 12 | 22 | 24.00 | 24 | -2 | 2 | 13 (NEW) | | 23.67 | 24 | 0.93 | 3.96 | b) Use exponential smoothing with an α of 0.6 to estimate the month-in-advance forecast of demand for months 4-12 and generate a forecast for the first month of next year. Calculate the Mean Error and the Mean Absolute Error Month | Demand | ESF | ESF(arredondado) | ME | MAE | 1 | 20 | - | - | - | - | 2 | 18 | - | - | - | - | 3 | 21 | 21.00 | 21 | - | - | 4 | 25 | 21.00 | 21 | 4 | 4 | 5 | 24 | 23.40 | 24 | 1 | 1 | 6 | 27 | 23.76 | 24 | 3 | 3 | 7 | 22 | 25.70 | 26 | -4 | 4 | 8 | 30 | 23.48 | 24 | 7 | 7 | 9 | 23 | 27.39 | 28 | -4 | 4 | 10 | 20 | 24.76 | 25 | -5 | 5 | 11 | 29 | 21.90 | 22 | 7 | 7 | 12 | 22 | 26.16 | 27 | -4 | 4 | 13 (NEW) | | 23.66 | 24 | 0.49 | 4.27 |

c) Compare the Mean Error and the Mean Absolute Error you obtained in (a) and (b) above. Based on these error calculations, which of the two forecast methods would you recommend and why? As was discussed in question 2, the forecast has to be close enough to the reality to be a good and help the demand management in allocation of resources, production, etc. Both forecasting have similar results, but not equal. To be a good forecast, the forecast has to have ME and MAE as close to zero as possible. So, the ME tell us to choose the exponential smoothing and the MAE tell us to choose the Moving Average. The Mean Error show that the exponential smoothing compensate the higher forecasts with lower forecasts(and vice-versa) more than the three-months moving average does. However, the MAE tell us that the three-months moving average has lower errors magnitude than the exponential smoothing. Analyzing the month by month error, we can see that the exponential smoothing have the highest error (in month 11, the forecast exceed the demand in 7 units) and, as was told by the MAE, it has a higher error magnitude. So, in order to choose the more accurate forecast method, I would recommend the three-months moving average.

Reference Vollman, T. E., Berry, W., L., Whybark, D. C., Jacobs, F. R., “Manufacturing Planning & Control for Supply Chain Management”, McGraw-Hill, 2005.