How to properly define a function that uses internal dataset - r

Question

I would like to create some R functions and combine them in a small R package for an university course. Those functions will take input values (tree diameter at breast height) and estimate an output parameter (biomass) based on a function with parameters published in an article. Those parameters depend on the tree species and the functions are of this form (just a simplified example):
fun <- function(diameter, species){
parameter <- dataframe$parameter[which(dataframe$species == species)]
mass <- parameter*diameter
return(mass)
}
Hence, the function will have to look up those parameters in a table (named dataframe in this example).
Now I wondered what the best practice would be, to implement this. I have the data as Excel tables and I can load them as data.frames in R. I could use dput() and paste the output into my functions, so they contain the data to look up the values. However, this approach is probably not the most efficient one? It will make the functions pretty much unreadable.
Creating the data.frame as a global variable in the user's environment is probably also not best practice?
So, I wondered, how data sets should be included in R functions. Unfortunately, I was too stupid to google a solution (searching for terms such as "data.frame" or "function" obviously leads to a lot of stuff not related to my question; I have no useful keywords to start with). I hope this question is not a duplicate (the suggestions show very different questions).

Answer 1

The question is whether the data are to be user-visible as test data set. Then the right place for the data frame is a text file in the /data subdirectory of the package. This is the standard method. However, if the data are just parameters, one can define a local environment in the package and then place the the data in an .R script together with the function /Rfolder of the package.
You may have a look at the source code of package marelac, a package for aquatic sciences that uses both techniques.
Finally one can store it in R/sysdata.R as shown in the link provided by #Gabor Grothendiek.
Briefly the method that was used in marelac: First, there is a file aaa.R that defines a new environment as follows:
.marelac <- new.env()
The file name aaa.R ensures that this is loaded first, the dot that it is a "hidden" variable.
Then a function can make use of it:
## -----------------------------------------------------------------------------
## Seawater Composition
## function taken from package marelac, license: GPL >= 2
## -----------------------------------------------------------------------------
.marelac$sw_comp <- c(Na = 0.3065958, Mg = 0.0365055, Ca = 0.0117186,
K = 0.0113495, Sr = 0.0002260, Cl = 0.5503396, SO4 = 0.0771319,
HCO3 = 0.0029805, Br = 0.0019134, CO3 = 0.0004078, BOH4 = 0.0002259,
F = 0.0000369, OH = 0.0000038, BOH3 = 0.0005527, CO2 = 0.0000121)
sw_comp <- function(species = c("Na", "Mg", "Ca", "K", "Sr", "Cl", "SO4", "HCO3",
"Br", "CO3", "BOH4", "F", "OH", "BOH3", "CO2")) {
species <- match.arg(species, several.ok = TRUE)
.marelac$sw_comp[species]
}
Here sw_comp is the standard composition of seawater. This is the smallest (almost trivial) function with this technique and there are some others, e.g. gas_solubilityor diffcoeff.