plotting in for loop with R using package eHOF - r

Question

I want to run a number (~100) of eHOF models using the R package eHOF, produce the graphs that this package can make of the models, and save jpeg files of each one. I am trying to use a for loop to accomplish this quickly, but I am not able to get it to make the graphs. I don't see anything in the R studio plots window, and I produce jpeg files that have nothing in them (as an aside problem, I am not producing the names for the jpeg files correctly in the loop).
To produce these plots, outside a loop there is no problem, if for example I call my model modSP<-HOF(Sp, ...) then using plot(ModSp) produces the desired graph. But within the loop, nothing is produced, and I get several error messages of the sort:
1: In xy.coords(x, y, xlabel, ylabel, log) : NAs introduced by coercion
2: In plot.window(...) : "boxp" is not a graphical parameter
3: In plot.window(...) : "las.h" is not a graphical parameter
4: In plot.window(...) : "onlybest" is not a graphical parameter
5: In plot.window(...) : "para" is not a graphical parameter
Background: I am using R version 3.1.0 (2014-04-10) in R studio, and package eHOF in windows 7.
My code is as follows:
species<- read.csv("F:/Thesis_projects/Chapter4_climateChange/HOF/259species.csv")
species<-as.data.frame(species)
enviro<-read.csv("F:/Thesis_projects/Chapter4_climateChange/HOF/EnvironmentalData.csv")
enviro<-as.data.frame(enviro)
species_enviro<-merge(enviro, species, all.x=FALSE)
HOF_Sp<-species_enviro[,23:25]
GDD<-species_enviro[,19]
library(eHOF)
SpeciesCodes<-c("ACPE","ACRU2","ACSP2")
Modx<-NULL
for (Spp in seq_along(SpeciesCodes)){
Modx[[Spp]]<-HOF(HOF_Sp[[Spp]],GDD, M=1,family=binomial, bootstrap=2, freq.limit = 100)
jpeg(filename = (paste(("GDD_responsecurve_",SpeciesCodes[[Spp]],".jpg"),sep="")),
width =8.3, height = 8.3, units = "cm", pointsize = 8, bg="white", res = 800)
print(plot((paste(c(Modx[[Spp]]))), boxp = TRUE,
las.h = 1, onlybest = TRUE, para = TRUE,
gam.se = FALSE, newdata = NULL, lwd=1, leg = TRUE, add=FALSE,
xlabel="Growing degree days", ylab="Probability"))
dev.off()
}
My Data looks like this:
> head(GDD)
[1] 996.1681 996.1681 962.0662 962.0662 945.7007 945.7007
(there are lots of 1's in the species data too, just not in the first few rows).
> head(HOF_Sp)
ACMI2 ACPA ACPE
1 0 0 0
2 0 0 0
3 0 0 0
4 0 0 0
5 0 0 0
6 0 0 0
Any advice at all would be very appreciated! I think it is an issue with the plot function in the loop being mistaken for the generic plot function of R. If I didn't provide enough information, I will be happy to edit my question.

Answer 1

There are several things going on here:
Your SpeciesCodes vector has codes different from the column names in HOF_Sp.
seq_along(...) returns the index of each element in it's argument, not the element itself.
The plot method for HOF is invoked when an object of class HOF is passed to plot(...). But you are passing paste(c(HOF)) which is incomprehensible...
Your example is not reproducible because it does not run with the data you provided (did you try to run it??). Specifically, the sample of HOF_Sp is degenerate because there are no non-zero values.
It's impossible for me to test this code because of (4) above, but try this:
SpeciesCodes <- c("ACPE","ACRU2","ACSP2")
for (Spp in SpeciesCodes) {
model <- HOF(HOF_Sp[[Spp]],GDD, M=1,family=binomial, bootstrap=2, freq.limit = 100)
jpeg(filename = (paste(("GDD_responsecurve_",Spp,".jpg"),sep="")),
width =8.3, height = 8.3, units = "cm", pointsize = 8, bg="white", res = 800)
plot(model, boxp = TRUE,
las.h = 1, onlybest = TRUE, para = TRUE,
gam.se = FALSE, newdata = NULL, lwd=1, leg = TRUE, add=FALSE,
xlabel="Growing degree days", ylab="Probability")
dev.off()
}
Note that this invokes the vector version of HOF(...), which I gather is what you want.
This does not solve the problem in (1) above (species codes do not match columns names), but other than that it should work.