USTobacco(FitAR) | R Documentation |
Annual U.S. tobacco production, 1871-1984, in millions of pounds.
data(USTobacco)
The format is: Time-Series [1:114] from 1871 to 1984: 327 385 382 217 609 466 621 455 472 469 ... - attr(*, "title")= chr "Tobacco production,US, 1871-1984"
Wei (2006, p.120, Example 6.6) fits an ARIMA(0,1,1) to the logarithms. But a more accurate Box-Cox analysis indicates a square-root transformation should be used.
Wei, W.W.S. (2006, Series W6, p.570), Time Series Analysis: Univariate and Multivariate Methods. 2nd Ed., New York: Addison-Wesley.
#A plot of the series we see the variance is increasing with level. #From the acf of the first differences an ARIMA(0,1,1) is suggested. data(USTobacco) # layout(matrix(c(1,2,1,2),ncol=2)) plot(USTobacco) lines(lowess(time(USTobacco), USTobacco), lwd=2, col="blue") acf(diff(USTobacco, differences=1))