6
votes

I have one data.table with 1M rows and 2 columns

Dummy data:

require(data.table)
ID <- c(1,2,3)
variable <- c("a,b","a,c","c,d")
dt <- data.table(ID,variable)
dt
> dt
ID variable
1      a,b
2      a,c
3      c,d

Now I want to collapse the column "variable" into different rows by "ID", just as the "melt" function in reshape2 or melt.data.table in data.table

Here's what I want:

ID variable
1  a
1  b
2  a
2  c
3  c
3  d 

PS: Given the desired results, I know how to do the reverse step.

dt2 <- data.table(ID = c(1,1,2,2,3,3), variable = c("a","b","a","c","c","d"))
dt3 <- dt2[, list(variables = paste(variable, collapse = ",")), by = ID]

Any tips or suggestions?

1
dt[, strsplit(variable, ","), by=ID]shadow
Look here.lukeA

1 Answers

5
votes

Since strsplit is vectorised, and that's going to be the time consuming operation here, I'd avoid using it on each group. Instead, one could first split on the , on the entire column and then reconstruct the data.table as follows:

var = strsplit(dt$variable, ",", fixed=TRUE)
len = vapply(var, length, 0L)
ans = data.table(ID=rep(dt$ID, len), variable=unlist(var))

#    ID variable
# 1:  1        a
# 2:  1        b
# 3:  2        a
# 4:  2        c
# 5:  3        c
# 6:  3        d