The article by Blikman states: The third parameter was the variation of bout length (coefficient of variation). This outcome was obtained after dividing the SD by the mean bout length (min) from the lognormal transformed data.
which means that we need to do instead:
sd(log(DurationPA)) / mean(log(DurationPA))
I think fragmentation metric CoV is not calculated correctly. https://github.com/wadpac/GGIR/blob/master/R/g.fragmentation.R#L200-201
The article by Blikman states: The third parameter was the variation of bout length (coefficient of variation). This outcome was obtained after dividing the SD by the mean bout length (min) from the lognormal transformed data.
which means that we need to do instead:
sd(log(DurationPA)) / mean(log(DurationPA))