[R] Differencing with auto.arima and xreg

[Kim H]

I am having some issues with differencing using auto.arima when also
specifying an xreg dataframe.

The xreg dataframe contains dummy variables that specify time periods that
had a promotion running.

When I model diff(y) with order (1,0,1), the coefficients for these dummy
variables are very different than when I model y with order=(1,1,1).  I
think when modeling diff(y) the coefficients represent the % change from
week to week (since the response is first logged) and I may be having issues
trying to run a dummy variable over consecutive periods since the promotion
is probably not causing increases in % changes at each time period.  But
when modeling y with order (1,1,1), it allows me to use the dummy variable
over consecutive periods and appears meaningful.

Does anyone have any insight as to why this is happening?
-- 
View this message in context: http://r.789695.n4.nabble.com/Differencing-with-auto-arima-and-xreg-tp2279874p2279874.html
Sent from the R help mailing list archive at Nabble.com.