search

kylebaron / mrgsim.sa

https://kylebaron.github.io/mrgsim.sa
6 stars 0 forks source link

questions about mrgsim.sa #4

Open kevin8897 opened 1 year ago

kevin8897 commented 1 year ago

what dose "eps" mean in the code, parameter change value extent or a ratio? how the sensitivity indices was calculated, and how the parameters changed in default?

Examples mod <- mrgsolve::house(delta=0.1)

par <- "CL,VC,KA"

var <- "CP"

dose <- ev(amt = 100)

fun <- function(mod, ...) mrgsolve::mrgsim_e(mod, dose, output="df")

out <- lsa(mod, par, var, fun)

kylebaron commented 1 year ago

Yeah ... it's a fractional change in the parameter value.

The source code is here: https://github.com/kylebaron/mrgsim.sa/blob/main/R/lsa.R#L85C1-L86

I need to write a vignette about this or get it in the documentation so you don't have to read the code for this.

kevin8897 commented 1 year ago

Yeah ... it's a fractional change in the parameter value.

The source code is here: https://github.com/kylebaron/mrgsim.sa/blob/main/R/lsa.R#L85C1-L86

I need to write a vignette about this or get it in the documentation so you don't have to read the code for this.

Hi, kylebaron, thanks a lot for your reply. so the default eps value is 1e-08?
and based on the output, dose lsa only run a single side analysis? in the source code, dose dvalue ="sens" in the output file? dvalue <- function(sim,ref,scale) { (sim[["dv_value"]] - ref[["dv_value"]])/scale}, what is scale? is it baseline ref[["dv_value"]]?

sorry for so many questiones, looking forward to the vignette, thanks!