subseting in a for loop - r

Question

My dataset has 34,000 rows and 353 columns. One column is location and it has 11,000 unique values. I want to subset the dataset within a for loop. I can do this by creating a new data frame for each subset, but I want the subsets to form a single data frame. I have included a sample dataset below
structure(list(X = structure(c(1L, 1L, 1L, 1L, 3L, 3L, 3L, 2L,
3L), .Label = c("Car", "DOG", "House"), class = "factor"), Y = c(20L,
20L, 20L, 20L, 410L, 410L, 410L, 410L, 60L), Z = structure(c(1L,
3L, 8L, 1L, 7L, 5L, 2L, 4L, 6L), .Label = c("ARGENTINA", "BERLIN GERMANY",
"BUENOS AIRES ARGENTINA", "DUBLIN IRELAND", "FROM AUSTRIA", "GERMANY",
"IN TRANSIT FROM GERMANY", "RIVER PLATE ARGENTINA"), class = "factor"),
K = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L), .Label = "A", class = "factor")),
.Names = c("X", "Y", "Z", "K"), class = "data.frame", row.names = c(NA, -9L))
I can use the following code to create new data frames
l=c("ARGENTINA","IRELAND")
for(i in l){
assign(paste("newdata",i,sep=""),
subset(TESTL[which(grepl(i,TESTL$Z)&
!grepl("IN TRANSIT",TESTL$Z)&!grepl("FROM",TESTL$Z)),],
select=c("X","Y","Z")))}
However I want to create a single new dataframe to hold all the subsets. I have tried the following code
d<-data.frame()
for(i in l){d<-rbind(d,c(
subset(TESTL[which(grepl(i,TESTL$Z) & !grepl("IN TRANSIT",TESTL$Z)
& !grepl("FROM",TESTL$Z)),],
select=c("X","Y","Z")))}
I get the following errors
Warning messages:
1: In `[<-.factor`(`*tmp*`, ri, value = "DOG") :
invalid factor level, NA generated
2: In `[<-.factor`(`*tmp*`, ri, value = "DUBLIN IRELAND") :
invalid factor level, NA generated
I have attempted to convert the factors to characters with no success. Any help appreciated

Answer 1

I think you are making your life rather difficult by using assign here and trying to store the subsets in separate data frames. Try something more like this:
l <- c("ARGENTINA","IRELAND")
res <- setNames(vector("list",length(l)),l)
for (i in seq_along(l)){
res[[i]] <- dat[grepl(l[i],dat$Z) & !grepl("IN TRANSIT",dat$Z) & !grepl("FROM",dat$Z),c("X","Y","Z")]
}
> res
$ARGENTINA
X Y Z
1 Car 20 ARGENTINA
2 Car 20 BUENOS AIRES ARGENTINA
3 Car 20 RIVER PLATE ARGENTINA
4 Car 20 ARGENTINA
$IRELAND
X Y Z
8 DOG 410 DUBLIN IRELAND
> do.call("rbind",res)
X Y Z
ARGENTINA.1 Car 20 ARGENTINA
ARGENTINA.2 Car 20 BUENOS AIRES ARGENTINA
ARGENTINA.3 Car 20 RIVER PLATE ARGENTINA
ARGENTINA.4 Car 20 ARGENTINA
IRELAND DOG 410 DUBLIN IRELAND

Answer 2

The warnings is becouse at first iteration of a loop (ARGENTINA) it introduces factors variables X and Z, and on the second indtroduce IRELAND with another factor levels. So:
First you should change a classes of your vaiables n TESTL:
for (i in names(TESTL) [grep ("factor", sapply (TESTL, class))]) {
TESTL[[i]] <- as.character (TESTL[[i]])
}
Then it will work with the next code:
d <- data.frame(stringsAsFactors=F)
for(i in l){d <- rbind(d,
TESTL [grepl(i,TESTL$Z) & !grepl("FROM|IN TRANSIT", TESTL$Z), c("X", "Y", "Z")])}