When in interactive mode, exploring aggregations in data.table, I may do dozens, or hundreds of experiments. The default column names after aggregations (V1, V2, etc) obviously aren't very informative, after writing the minimal amount of code needed to generate the aggregations. Often, I'd be happier with the default column name of a simple aggregation of one column, like a mean or sum, to be just the name of the underlying variable.
All the extra column name typing gets tiring, and I want to avoid this.
Is there any easy way to do this in data.table?
e.g. a simplified example to demonstrate if it's not clear:
DT = data.table(x =rep(c("b","a","c"),each=3), y_a_long_name=c(1,3,6), v_a_long_name=1:9)
DT[, .(sum(v_a_long_name), mean(y_a_long_name)), by = x]
# x V1 V2
# 1: b 6 3.333333
# 2: a 15 3.333333
# 3: c 24 3.333333
When you start working with several columns, using different aggregation type functions, the above labeling, V1, V2, isn't helpful.
All the extra typing of repeating the names is tedious, but I'd like to see something like this:
DT[, .(v_a_long_name = sum(v_a_long_name), y_a_long_name = mean(y_a_long_name)), by = x]
# x v_a_long_name y_a_long_name
# 1: b 6 3.333333
# 2: a 15 3.333333
# 3: c 24 3.333333
while typing as minimal as possible. e.g. it would be ideal if
DT[, .(sum(v_a_long_name), mean(y_a_long_name)), by = x]
printed this by default:
# x v_a_long_name y_a_long_name
# 1: b 6 3.333333
# 2: a 15 3.333333
# 3: c 24 3.333333
GForcefrom optimizing your queries. Does your IDE not have auto-complete? If you're just playing around, I don't see much penalty to using names likeV1,V5. If this is not ephemeral, writing explicit code will help others (including future-you) better understand your code. Choosing to eliminatesumandmeanfrom the name of the aggregated variable strikes me as troublesome / begging for confusion/errors down the line. – MichaelChiricov_a_long_name_sumandy_a_long_name_mean) – MichaelChirico