Subset using column name defined using paste0 and a variable - r

Question

I have a data.table like this:
dt <- data.table(asset=c("x1","x2","x3","x4","x5"),
min_s1=c(.1,NA,NA,.1,NA),
min_s2=c(NA,.5,.5,NA,NA),
min_s3=c(.15,NA,NA,NA,.15))
I can manually subset on the NA values as follows which gives me the output I want:
empty1 <- dt[is.na(min_s1)]
empty2 <- dt[is.na(min_s2)]
empty3 <- dt[is.na(min_s3)]
But, what I really need to do is subset dynamically using the column name and also name the result incorporating the i variable in a loop. The loop is important because this will ultimately be used in a parallel computing script. I would like something like this (it doesn't work; just showing what I am looking for):
foreach (i in 1:3) %do% {
empty(i) <- dt[is.na(min_s(i))]
}
I tried using the following as well as many of its variations to no avail:
paste0("empty",i) <- dt[is.na(paste0("min_s",i))]
Any ideas how I could accomplish this?
I use a Windows 7 pc.
Thanks.

Answer 1

We can just loop through the 'min' columns using lapply and subset the dataset
lapply(dt[,-1, with =FALSE], function(x) dt[is.na(x)])
#$min_s1
# asset min_s1 min_s2 min_s3
#1: x2 NA 0.5 NA
#2: x3 NA 0.5 NA
#3: x5 NA NA 0.15
#$min_s2
# asset min_s1 min_s2 min_s3
#1: x1 0.1 NA 0.15
#2: x4 0.1 NA NA
#3: x5 NA NA 0.15
#$min_s3
# asset min_s1 min_s2 min_s3
#1: x2 NA 0.5 NA
#2: x3 NA 0.5 NA
#3: x4 0.1 NA NA

Answer 2

I hope I've understood your question correctly, so try the following:
dt <- data.table(asset=c("x1","x2","x3","x4","x5"),
min_s1=c(.1,NA,NA,.1,NA),
min_s2=c(NA,.5,.5,NA,NA),
min_s3=c(.15,NA,NA,NA,.15))
vec_store <- c()
empty <- list()
names <- names(dt)[!grepl("asset", names(dt))]
for(i in names){
vec_store <- dt[is.na(dt[,get(i)])]
empty[[paste0(i)]] <- vec_store
}
This gives you:
> empty
$min_s1
asset min_s1 min_s2 min_s3
1: x2 NA 0.5 NA
2: x3 NA 0.5 NA
3: x5 NA NA 0.15
$min_s2
asset min_s1 min_s2 min_s3
1: x1 0.1 NA 0.15
2: x4 0.1 NA NA
3: x5 NA NA 0.15
$min_s3
asset min_s1 min_s2 min_s3
1: x2 NA 0.5 NA
2: x3 NA 0.5 NA
3: x4 0.1 NA NA
In your code, you can't use paste0(empty,i) to create objects, I find the easiest way to do this is to create an empty vector, or data.table, then store it in a list. Once it is in the list you can do operations on the list, or pull it out of the list individually. That is why i created the vec_store an empty vector, and empty an empty list.
Hope that helps.