dyn is an R package that facilitates the use of regression
using time series data with lags and diffs (known as dynamic
regression). It is a lightweight package that has no
facilities of its own but leverages off the various time
series and regression functions in R to make it easier to
use them together.
Its features include:
- many regression functions. It can be used with lm, glm,
loess, rq, randomForest, lqs, rlm and any other regression
functions that use model.frame and are written in the
style of lm.
- many time series classes. It can be used with ts, zooreg,
zoo, its, and irts time series classes. This covers
regular, weakly regular and irregular time series classes.
- missing values. Time series may have missing values including
internal missing values. Both na.omit and na.exclude are
supported.
- good citizen. It does not replace the regression
functions but rather works with them by providing new
methods to the standard R generics: model.frame, resid,
fitted, predict, update, anova and $.
- ease of use. dyn enables one to use the same regression
functions (lm, glm, etc.) using the same syntax one has
always used. Just preface the regression function name with
dyn$ and it is transformed into a regression function that
can handle time series:
dyn$lm( y ~ x + lag(x) + diff(w) ) # lm
dyn$loess( y ~ x + lag(x) + diff(w) ) # loess
- modular. dyn can be used with any regression function that
uses model.frame and is written in the style of lm. Additional
classes can be added to dyn simply by adding new methods. dyn
is modular so such updates can be made without changing dyn,
itself.
- documentation. It includes a help page and six demos.
?dyn # help file
demo() # look under dyn for list of demos
demo("dyn-rq") # runs indicated dyn demo
The package is available on CRAN. Comments/questions welcome.
_______________________________________________
R-packages mailing list
R-packages at stat.math.ethz.ch
https://stat.ethz.ch/mailman/listinfo/r-packages
