brown, durbin , evans ( 1975 )

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

brown, durbin , evans ( 1975 )

mark leeds-4
Does anyone know where
I can get  R code for plotting
the Brown , Durbin
and Evans cumsum
procedure ( 1975 ) ?
 
I wrote my own code but
I am a little worried
that my confiodence bands
may not be correct ( I find the formula
in the original paper confusing and S+Finmetrics
has a formula but that formula implies that
there should be 4 lines as far as I can tell ) so
I would like to see someone
else's code also if it's out
there ? Thanks.
 
                      Mark
 
P.S : Strucchange looks like a great package but
it would take me way longer than I
have to only partially understand the theory behind
what is going on there.
 
 
 
 
 
 


**********************************************************************
This email and any files transmitted with it are confidentia...{{dropped}}

______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Reply | Threaded
Open this post in threaded view
|

Re: brown, durbin , evans ( 1975 )

Achim Zeileis
Mark:

> Does anyone know where
> I can get  R code for plotting
> the Brown , Durbin
> and Evans cumsum
> procedure ( 1975 ) ?

Via
library("strucchange")

> P.S : Strucchange looks like a great package but
> it would take me way longer than I
> have to only partially understand the theory behind
> what is going on there.

Come on, it's not that difficult, e.g., for detecting the mean shift in
  plot(Nile)
you can apply BDE's Recursive CUSUM test via
  rcus <- efp(Nile ~ 1, type = "Rec-CUSUM")
  plot(rcus)
which shows quite clearly the structural change that occured in 1898 after
the Ashwan dam was built. You can get test statistic and p-value via
  sctest(rcus)

See the package vignette and the man pages for more examples.

As I already told you privately, a lot of research was done after BDE 1975
and the test is not the most suitable in various situations. For example,
it can perform quite badly in the presence of multiple shifts or shifts
that occur late in the sample period.

Best,
Z

______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html