[R] weird things in seqMK() and sqmk()

[Nick Wray]

Hello  The other say I posted a question about altering confidence limits
in the seqMK() function  in the pheno package.  I didn't get any bites but
in any case I've been trying to use this function.  It seems to me to give
very odd results - sometimes identifying 8 or 9 changepoints in a
relatively short vector of maybe 40 elements, and at other times not giving
any changepoints at all, despite visually there appearing to be
changepoints and other changepoint algorithms (pettitt etc) saying that
there are other changepoints.  Apart from the cran vignette there doesn't
seem to be anything on this function

Has anyone else used the seqMK() function and if so, have they noticed any
weird behaviour with it?

I've also lookedf at sqmk() in trendchange function.  This also seems to
give odd results, and not ones consistent with seqMK() results either

Again has anyone used these and do they know of any sites with further ifo,
as it seems v sparse about these two functions
Thanks Nick Wray

	[[alternative HTML version deleted]]