artfima (version 1.5)

seriesa: Series A from Box and Jenkins

Description

Chemical process concentration readings every two hours is comprised of 197 consecutive observations. Box and Jenkins fit ARMA(1,1) and ARIMA(0,1,1) to this data.

Usage

data("seriesa")

Format

The format is: Time-Series [1:197] from 1 to 197: 17 16.6 16.3 16.1 17.1 16.9 16.8 17.4 17.1 17 ...

Source

listed in Box and Jenkins book

References

Box and Jenkins (1970). Time Series Analysis: Forecasting and Control.

Examples

data(seriesa)
#compare ARMA(1,1) models and timings
system.time(arima(seriesa, order=c(1,0,1)))
system.time(artfima(seriesa, arimaOrder=c(1,0,1)))
#Remark: there is a slight difference due to the fact that arima() 
#uses the exact MLE for the mean parameter whereas artfima() uses
#the sample average. In practice, the difference is almost negible.
#
#Find AIC/BIC 3 best models. Takes about 15 sec
## Not run: 
# system.time(ans <- bestModels(seriesa, nbest=3))
# summary(ans) #summary provides plausibility as well as scores
# ## End(Not run)