You have been hired to help Mary with her forecasting problem. Write her a memo that summarizes her efforts to date and that recommends a course of action to her. Be sure your memo addresses the following issues: Should Mary accept the forecasts from one of her current procedures? If so, how should she "sell" the forecasts to management? Given the somewhat inconsistent character of her time series, should Mary search for another forecasting technique? Should she use one of her current methods but with a modified data set?
Mary Beasley is not very happy. She has tried several time series methods to develop forecasts for the monthly totals of billable visits to Medical Oncology (see Cases 4-7, 5-8, 8-9, and 9-10). She has had the greatest success with an ARIMA model, although the residual autocorrelations as judged by the Ljung-Box Q statistic (see Equation 9.5) seem to be collectively large.