Seasonal Series: Forecasting and Decomposition
INSTRUCTIONS: Use MINITAB 17 to complete the following questions. You are to work independently on this assignment. All data sets for this assignment have been provided. Submit completed assignment that includes relevant MINITAB output and comments (make sure to comment on your results when asked) as ONE WORDdocument.
1. Use the data set NAPAVALLEY.MTW. These data represent monthly wine sales for Northern Napa Valley Winery, Inc. from January 1988 through August 1996. The forecast variable is Monthly Wine Sales (C1).
(a) Fit additive and multiplicative decomposition models (with trend and seasonality) to these data(Monthly Wine Sales).
(b) Fit the Winter's triple exponential smoothing model (both additive and multiplicative) to these data(Monthly Wine Sales). Use the default values for thesmoothing constants.
(c) Based on the fitted accuracy measures, which model is best? .
2. Use the data set MurphyBrothersFurniture.MTW. These data represent monthly sales ($ billions) for all retail stores during the period from 2001 to 2013. Holdout the data for the last 6 months (i.e., do not use these data in the model fitting stage). Simply cut and paste the last six observations from C1 into C2. You will then use these observations to assess forecast accuracy. This means that you will use 150 observations for model fitting.
(a) Fit additive decomposition with trend and seasonal to these data. Forecast for the last 6 months. Store these forecasts (go to Storage and check Forecasts; the forecasts generated will then be stored in the next available column, in this case C3).
(b) Fit multiplicative decomposition with trend and seasonal to these data. Forecast for the last 6 months. Store these forecasts (go to Storage and check Forecasts; the forecasts generated will then be stored in the next available column, in this case C6).
(c) Which model fits the data better? Justify your answer.
(d) Calculate both MAE and MSE to measure forecast accuracy over the holdout sample (the last 6 months) for using forecasts generated from additive and multiplicative decomposition. Based on these results, which model forecasts better?
3. Use the data set CallCenter.MTW. These data represent monthly call volume for an existing product (Source:Quality Engineering 24 (2012): 386-399).
(a) Prepare a time series plot. What time series components are evident?
(b) Fit the Winter's triple exponential smoothing model (both additive and multiplicative) to these data. Use the default values for thesmoothing constants. Which fits better?
Attachment:- Data.rar