[R-pkgs] New forecasting bundle of packages

[Rob Hyndman]

v1.0 of the forecasting bundle of packages is now on CRAN and will 
propagate to mirrors shortly.

The forecasting bundle of R packages provides new forecasting methods, 
and graphical tools for displaying and analysing forecasts. It comprises 
the following packages:

     * forecast: Functions and methods for forecasting.

     * fma: All data sets from Makridakis, Wheelwright and Hyndman 
(1998)         Forecasting: methods and applications, Wiley & Sons: New 
York.

     * Mcomp: All data from the M1 and M3 forecast competitions.

Key features:

* a "forecast" method and class which can be applied to Arima, StructTS, 
HoltWinters and other time series models. This is preferred to predict() 
as it provides output in a consistent format (the "forecast" class) that 
can be used by other functions.

* automatic univariate time series forecasting based on exponential 
smoothing state space models. This is much more general and flexible 
than HoltWinters().

* automatic ARIMA forecasting based on minimizing the AIC or BIC.

* several new forecasting methods and time series graphics.

Some features of the forecast package were the subject of my talk at 
UseR! in Vienna in June. Slides of the talk are at 
http://www.robhyndman.info/talks/Hyndman_UseR.pdf

Anyone who has been using earlier versions of the packages from my web 
pages should check out the list of changes at 
http://www.robhyndman.info/Rlibrary/forecast/

__________________________________________________
Professor Rob J Hyndman
Department of Econometrics & Business Statistics,
Monash University, VIC 3800, Australia
http://www.robhyndman.info/