It seems like simulated data should generally go through the equivalent transformations as observed data. For that reason, I think that it makes sense to use quantile_cens() instead of the regular quantile() function. The benefit would be that simulated data would have the same censoring at the LLOQ or ULOQ as observed.
It seems like simulated data should generally go through the equivalent transformations as observed data. For that reason, I think that it makes sense to use
quantile_cens()
instead of the regularquantile()
function. The benefit would be that simulated data would have the same censoring at the LLOQ or ULOQ as observed.What do you think?
Relevant code is in
calc_vpc_continuous.R
:https://github.com/billdenney/vpc/blob/ccbc2474ba5e47cd48db84201338fd0749ab5a1f/R/calc_vpc_continuous.R#L16-L24