Any possibility of adding na.rm=FALSE argument to the genDataDensity and addDataDensity functions?

[vlulla]

The below example demonstrates a small issue i ran into when i was trying to generate data from an arbitrary distribution using the genDataDensity function:

R> library("NHANES")
R> library("simstudy")
R> x <- NHANES$Height
R> d <- genDataDensity(n=500L,x,varname="ht")
Error in density.default(dataDist, n = 10000) :
  'x' contains missing values
R> d <- genDataDensity(n=500L,x[!is.na(x)],varname="ht") ## workaround
R> ## d <- genDataDensity(n=500L,x,varname="ht",na.rm=TRUE) ## would be nice...

Since na.rm=FALSE is also the default for stats::density, I understand why you do not include na.rm argument for these functions. Can I please make a request to add na.rm argument to the genDataDensity and addDataDensity functions? I will benefit from having this pass-through argument as a part of these function calls. And, since i know a workaround I can definitely generate the data without this argument. Therefore, whatever you decide is fine with me.

Thanks for a great package! I love using simstudy!

[kgoldfeld]

Great suggestion. It wasn't an intentional decision - it was just an oversight. The change is definitely doable. I had another thought - do you think it would be useful to generate missingness at the same rate of the source data if the user does not specify na.rm=TRUE?

[vlulla]

What a great idea! I hadn't thought of this but I anticipate that generating missingness at the same rate of the source data would be useful. The one application that immediately came to my mind was simulating presence/absence data for ecological niche modeling and species distribution modeling! I'll have to discuss this with my ecologist friends. Thank you for sparking this interesting question!

[kgoldfeld]

@vlulla I implemented the change - should be available in the development version. Let me know if it works for you.

[vlulla]

Thank you!