Question: 1. Assume that the following demands vary according a four-period seasonal cycle:
a. Compute the seasonal indexes using the average demand in each cycle as the base.
b. Compute the seasonal indexes using regression estimates as the base.
c. How do the answers for parts a and b differ? How would you explain the difference?
2. Use the Excel spreadsheet that accompanies this supplement to evaluate different forecasting models using the ice cream sales data.
a. Which model, time series regression, causal regression using temperature, or trend enhanced exponential smoothing, gives better forecast accuracy? Report all bias and accuracy metrics.
b. What combination of parameters for the trend enhanced smoothing model gives the best results?
c. Calculate seasonally adjusted forecasts, first using the average demand as the base, then using the time series regression forecasts as the base, then using the causal regression forecasts as the base. Which model is better? Why?