[R] strucchange: breakpoints in inequally spaced data
Achim.Zeileis at wu-wien.ac.at
Mon Jul 19 11:22:18 CEST 2004
I already replied privately to Thomas about this issue; here is a brief
reply to the list, just in case anyone is interested:
> we want to identify breakpoints (different phases) in environmental
> data, algae cell counts of three years with intervals between 7 and 30
> days (N=40). We found that
> breakpoints(cells ~1)
> works great and identifies 5 very good breaks, however we are
> uncertain about these, because the data are unequally spaced. Is there
> a way to include the information about the measurement intervals, e.g.
> breakpoints(cells[-1] ~ diff(time))
None of the methods in strucchange require the data to be equally
spaced, so both models could lead to sensible estimates of the
breakpoints. It depends on the biological interpretation which of the
two models is to be preferred.
> Furthermore we wonder, why the OLS-CUSUM indicates several of these
> breakpoints but does not mark them as significant.
If there are mutliple breaks in opposite directions, the OLS-CUSUM test
might be a bad choice. MOSUM or ME techniques or F statistics might be
more suitable for detecting multiple breaks.
> Thanks in advance!
> Thomas P.
> R-help at stat.math.ethz.ch mailing list
> PLEASE do read the posting guide!
More information about the R-help