The percentages provided must be the same length as the number of bumps!

[Stephocke]

Hi guys, I got the following error:

"The percentages provided must be the same length as the number of bumps!"

as I utilize a custom relevance function and C.perc arguments. The documentation does not say anything about bumbs.
The connection between relevance function and bumbs is still not clear to me, because the error occurs when I change the following line:

from:
s <- boxplot.stats(data$y)$stats
rel <- matrix(0, ncol = 3, nrow = 0)
# line of interest
rel <- rbind(rel, c(0.9*s[1], 0.6, 0))
# end line of interest
rel <- rbind(rel, c(s[2], 1, 0))
rel <- rbind(rel, c(0.9*s[3],1,0))
rel <- rbind(rel, c(s[4],0,0))

into:
s <- boxplot.stats(data$y)$stats
rel <- matrix(0, ncol = 3, nrow = 0)
# line of interest
rel <- rbind(rel, c(0.9*s[1], 0.2, 0))
# end line of interest
rel <- rbind(rel, c(s[2], 1, 0))
rel <- rbind(rel, c(0.9*s[3],1,0))
rel <- rbind(rel, c(s[4],0,0))

So my original intention is to reduce the relevance of point 0.9*s[1] from the value 0.6 to anything else. However, WERCSRegress now counts a different number of classes. That makes no sense to me. Can you might give an explaination what's the issue here?

The number of bumbs seems somehow related to number of elements consisted by C.perp list argument.
However, the number of points in my relevance functions never equals the number of elements in C.perc.
The function takes offense at the value of relevance, which I should actually be able to determine freely.

set.seed(123)
test <-WERCSRegress(frml, data, 
                    rel = rel, thr.rel = 0.6, O = 0.8, U = 0.0,
                    C.perc = list(2.5,0.2))