The Forecast Process, Data Considerations, and Model Selection 89 c. What would be the best point estimate for the population class size? d. What is the 95 percent confidence interval for class size? What is the 90 percent confidence interval? Does the difference between these two make sense? 8. CoastCo Insurance, Inc., is interested in forecasting annual larceny thefts in the United States using the following data (c2p8): Larceny Larceny Year Thefts Year Thefts 1972 4, 151 1984 6,592 1973 4,348 1985 6,926 1974 5,263 1986 7,257 1975 5,978 1987 7,500 1976 6,271 1988 7,706 1977 5,906 1989 7,872 1978 5,983 1990 7,946 1979 6,578 1991 8, 142 1980 7, 137 1992 7,915 1981 7, 194 1993 7,821 1982 7,143 1994 7,876 1983 6,713 a. Prepare a time-series plot of these data. On the basis of this graph, do you think there is a trend in the data? Explain. b. Look at the autocorrelation structure of larceny thefts for lags of 1, 2, 3, 4, and 5. Do the autocorrelation coefficients fall quickly toward zero? Demonstrate that the criti- cal value for rx is 0.417. Explain what these results tell you about a trend in the data. c. On the basis of what is found in parts a and b, suggest a forecasting method from Table 2.1 that you think might be appropriate for this series