问题
Let DT be a data.table:
DT<-data.table(V1=factor(1:10),
V2=factor(1:10),
...
V9=factor(1:10),)
Is there a better/simpler method to do multicolumn factor conversion like this:
DT[,`:=`(
Vn1=as.numeric(V1),
Vn2=as.numeric(V2),
Vn3=as.numeric(V3),
Vn4=as.numeric(V4),
Vn5=as.numeric(V5),
Vn6=as.numeric(V6),
Vn7=as.numeric(V7),
Vn8=as.numeric(V8),
Vn9=as.numeric(V9)
)]
Column names are totally arbitrary.
回答1:
Yes, the most efficient would be probably to run set in a for loop
Set the desired columns to modify (you can chose all the names too using names(DT) instead)
cols <- c("V1", "V2", "V3")
Then just run the loop
for (j in cols) set(DT, i = NULL, j = j, value = as.numeric(DT[[j]]))
Or a bit less efficient but more readable way would be just (note the parenthesis around cols which evaluating the variable)
## if you chose all the names in DT, you don't need to specify the `.SDcols` parameter
DT[, (cols) := lapply(.SD, as.numeric), .SDcols = cols]
Both should be efficient even for a big data set. You can read some more about data.table basics here
Though beware of converting factors to numeric classes in such a way, see here for more details
来源：https://stackoverflow.com/questions/31695971/r-data-table-multi-column-coversion-by-names