Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.



Re: time series multiple regression
Posted:
Feb 1, 2012 9:18 PM


On Feb 1, 4:54 pm, Dick Startz <@> wrote: > >> In econometrics it's pretty standard to run a multiple regression > >> where the error terms are serially correlated. Any econometrics > >> package will handle this. > >> Dick Startz > > >Thanks Dick. Can you refer me to just one? (Preferably in R). I > >tried the Farnsworth package, but that does not appear to support > >multiple regression of time series. > > >Many thanks, > >Eric > > Unfortunately, I'm not an R user. (Could someone else chime in here?) > Two commercial econometrics packages that will do this are EViews and > Stata. (Also the packages TSP and RATS.) I'm pretty sure the open > source package gretl will handle this. > Dick Startz
Thanks Dick.
I'm now looking into Dynamic Linear Models, Functional Linear Models, Dynamic Systems Estimation and State Space models. While I've got some functioning code, unfortunately I don't know enough yet to understand the outputs. I can generate one step ahead predictions, but I was hoping to resimulate an entire time series with alternative scenarios of one independent variable.
This feels a little less trivial than a standard linear model.
Eric



