1. What does the initial plot of the data reveal?
2. How many degrees of differencing does it take to make the model stationary?
3. What does this tell you about the series?
4. Once the series is stationary, what appears to be happening with the data?
5. Looking at the ACF and PACF, what ARIMA model would be a likely starting point?
6. Using SPSS Expert Modeler, what model appears to provide the best fit to the data?
7. What percent of the total variation in the series is explained by the model?
8. What percent of the variance in the series is explained beyond the stationary mean?
9. What is the value of the Ljung-Box Q statistic?
10. What does this statistic tell us?
11. What is the smoothing parameter for the random or level component of the series?
12. What does this parameter tell us?
13. What is the smoothing parameter for the trend component?
14. What does this tell you about the forecast for the trend component?
15. What is the forecast for the number of subscribers for April 2004?
16. What is the 95% upper confidence level for this forecast?
17. What would the lower confidence level be for this month if we did a 99% confidence interval? )
18. What is the seasonal factor for April using this same model?
19. Which month has the smallest seasonal index?
20. Now, use the total number of subscribers as the time series. How many degrees of difference does it take to make the series stationary?
21. What does this tell you about the trend in the series for total subscribers?
22. Knowing the model that Expert Modeler provided for market 6, what is the equivalent ARIMA model for this series?
23. Comparing this ARIMA model for market 6 with the one that Expert Modeler provided for market 6, which one is better based on the Stationary R-squared measure of model fit?
24. Does the Normalized BIC comparison affirm that the Wintersâ€™ Additive model is better for market 6?
25. Does using the total number of subscribers as an independent variable improve the ARIMA model for market 6?