A national supplier of jet fuel is interested in forecasting its sales. These sales data are shown for the period from 1992Q1 to 2007Q4 (data in billions of gallons):

a. Convert these data to a time-series plot. What, if any, seasonal pattern do you see in the plot? Explain.

b. Deseasonalize the data by calculating the centered moving average, and plot the de-seasonalized data on the same graph used in part (a). Calculate the seasonal index for each quarter, and write a short explanation of why the results make sense.

c. Develop a trend for the data based on the centered moving averages, and plot that trend line on the graph developed in part (a). Compare the deseasonalized data (CMA) and the trend line. Does there appear to be a cyclical pattern to the data? Explain. d. Calculate the cycle factors and plot them on a separate time-series graph. Project the cycle factor ahead one year.

e. For the historical period, plot the values estimated by the time-series decomposition model along with the original data.

f. Make a forecast of sales for the four quarters of2008, and given the following actual data for that year, calculate the root-mean-squared error:

g. Develop two other forecasts of jet fuel sales with:

1. An exponential smoothing method; and

2. A regression model using just time and quarterly dummy variables. Compare the RMSE for the three models you have developed, and comment on what you like or dislike about each of the three models for this application.

Transcribed Image Text:

46733854068 39218340348 24958670394 33344445556 81653433293 77947776853 69216325731 33455555567 37915359985 69533769887 26850013608 33345555566 873221799 771030765 3508969441 3344444556 9 2 78901234567 99900000000 99900000000 11122222222