用data.table（R）中的ID变量将列替换为总和

Question

I'm trying to aggregate a certain subset of my variables according to an ID. I don't want to store the result as a new variable, because the sum will replace the old variables.

I'm looking for a simple way to do this using a data.table.

For now, I've got a workaround, and I'm hoping to simplify it if possible (ie one-line it):

sum_vars <- c("x1","x2","x4")
tempp <- dt[ , lapply(.SD, sum), by=ID, .SDcols=sum_vars]
dt[ , c(sum_vars) := NULL]
dt <- dt[tempp]
rm(tempp)

The problems I'm running into with one-lining (to get around creating that temporary variable) are:

tempp is a different size data frame than dt --all duplicates by ID are removed. So something like this doesn't work:

dt[ , sum_vars] <- dt[ , lapply(.SD, sum), by=ID, .SDcols=sum_vars]

Also, the following in-line merge creates new variables with .1 as a suffix (eg x1.1):

dt <- dt[dt[ , lapply(.SD, sum), by=ID, .SDcols=sum_vars]]

I want something like this to work, but it doesn't:

dt[ , .SD:=sum(.SD), by=ID, .SDcols=sum_vars]

But this just creates a variable named .SD

Minimialist data example

Start with

dt <- structure(list(ID = c(1L, 1L, 2L, 3L), x1 = c(1L, 1L, 1L, 1L), 
                     x2 = c(1L, 2L, 5L, 8L), x3 = c(1L, 3L, 6L, 9L), 
                     x4 = c(1L,  4L, 7L, 2L)), 
                .Names = c("ID", "x1", "x2", "x3", "x4"), 
                class = "data.frame", row.names = c(NA, -4L))
dt
#   ID x1 x2 x3 x4
# 1  1  1  1  1  1
# 2  1  1  2  3  4
# 3  2  1  5  6  7
# 4  3  1  8  9  2

end with

# ID x1 x2 x3 x4
# 1  2   3  4  5
# 1  2   3  4  5
# 2  1   5  6  7
# 3  1   8  9  2

Answer 1

See the data.table Reference Semantics vignette on GitHub:

Note that since we allow assignment by reference without quoting column names when there is only one column as explained in Section 2c, we can not do out_cols := lapply(.SD, max). That would result in adding one new column named out_col. Instead we should do either c(out_cols) or simply (out_cols). Wrapping the variable name with ( is enough to differentiate between the two cases.

You need to pass the appropriate vector of column names to the LHS of the call to :=

Therefore the following should work (replacing the values in the original dataset)

dt[,(sum_vars) := lapply(.SD, sum), by = ID, .SDcols = sum_vars]

If you wanted to preserve dt

dt_sum <- copy(dt)[,(sum_vars) := lapply(.SD, sum), by = ID, .SDcols = sum_vars]

Note that in both cases I have wrapped the vector of variable names ( sum_var on the LHS of := ) in () to force this to be evaluated (and not slmply create a column called sum var