Yankee Case Study
Comment on the forecasting system being used by Yankee.
What would be the smoothing forecast for Yankee for Month 1 of Year 5 using the four Month 1 data points in each of the last 4 years using an alpha of .3. (the demand for Month 1 of year 1 is the most historical data point.) Do this manually setting the very first forecast equal to the very first demand value)
What would be the smoothing forecast (using alpha = .3) for Month 1 of Year 5 using the 12 months given for Year 4? (Do this using POM for Windows software.) -- ONLY TYPE IN ACTUAL DATA POINTS INTO POM SOFTWARE --- LEAVE
FORECAST COLUMN FOR DATA ENTRY AS BLANK.
Repeat questions 2 and 3 above but use regression analysis with time as the independent variable. (You can do the regression runs using POM for Windows.) You need to do two separate regression forecasts.
From the above the techniques for just questions 2,3 & 4, used so far which one yields the best forecast in terms of MAD values BASED ON ACTUAL VERSUS FORECASTED VALUES. (You can pull MAD values of the POM for Windows software, but you are to find MAD for Question 2 manually, should you be asked to find it on an exam – keep in mind that POM for Windows drops out the very first error when using smoothing but keeps all errors when using regression -- so you are also to do this MAD calculation manually.)
For the best technique above using MAD as your criteria you are to develop a 95% confidence limits around the forecast of Month 1 for Year 5. (Formula for confidence limits -- high and low values around the forecast -- given in class lecture.) 7. Now what would be the best alpha to use to minimize MAD? ( USE 12 DATA POINTS IN QUESTION 3 FOR THIS PART). Depending on the version of POM for Windows you can have the computer generate the best alpha value for you -- put in 0 for alpha and computer will search through about 100 different alpha values. Otherwise you can try, say, ten different values of alpha ranging between >0 and
What would be the smoothing forecast for Yankee for Month 1 of Year 5 using the four Month 1 data points in each of the last 4 years using an alpha of .3. (the demand for Month 1 of year 1 is the most historical data point.) Do this manually setting the very first forecast equal to the very first demand value)
What would be the smoothing forecast (using alpha = .3) for Month 1 of Year 5 using the 12 months given for Year 4? (Do this using POM for Windows software.) -- ONLY TYPE IN ACTUAL DATA POINTS INTO POM SOFTWARE --- LEAVE
FORECAST COLUMN FOR DATA ENTRY AS BLANK.
Repeat questions 2 and 3 above but use regression analysis with time as the independent variable. (You can do the regression runs using POM for Windows.) You need to do two separate regression forecasts.
From the above the techniques for just questions 2,3 & 4, used so far which one yields the best forecast in terms of MAD values BASED ON ACTUAL VERSUS FORECASTED VALUES. (You can pull MAD values of the POM for Windows software, but you are to find MAD for Question 2 manually, should you be asked to find it on an exam – keep in mind that POM for Windows drops out the very first error when using smoothing but keeps all errors when using regression -- so you are also to do this MAD calculation manually.)
For the best technique above using MAD as your criteria you are to develop a 95% confidence limits around the forecast of Month 1 for Year 5. (Formula for confidence limits -- high and low values around the forecast -- given in class lecture.) 7. Now what would be the best alpha to use to minimize MAD? ( USE 12 DATA POINTS IN QUESTION 3 FOR THIS PART). Depending on the version of POM for Windows you can have the computer generate the best alpha value for you -- put in 0 for alpha and computer will search through about 100 different alpha values. Otherwise you can try, say, ten different values of alpha ranging between >0 and