unique can be optimized on keyed data.tables

Open
#2,947 12 comments 0 reactions 0 assignees View on GitHub

A pull request for this has already been merged.

  • #4386 by @jangorecki — merged

Assessment

Difficulty
4/5
Estimated time
3-5 days
Newbie friendliness
20/100
Issue type
Feature
Clarity
Mostly clear
Activity status
Stale
Tech stack
r
Domain
data, performance

Research direction

Start with the unique(DT$V1), DT[, unique(V1)], and DT[, TRUE, keyby = V1] entry points shown in the issue, then review linked pull request #4386. Done means unique on keyed data.tables reaches the keyby performance described by the benchmark, with the relevant behavior covered by the project’s tests.

Written by the indexing model from the issue text.

Description

enhancement

keyed tables are already known sorted, so finding unique values is much easier than it is in the general case.

Compare:

NN = 1e8

set.seed(13013)
# about 400 MB, if you're RAM-conscious
DT = data.table(sample(1e5, NN, TRUE), key = 'V1')

system.time(unique(DT$V1))
#    user  system elapsed 
#   1.354   0.415   1.798 

system.time(DT[ , unique(V1)])
#    user  system elapsed 
#   1.266   0.414   1.681 

system.time(DT[ , TRUE, keyby = V1])
#    user  system elapsed 
#   0.375   0.000   0.375 

It seems to me we should be able to match (or exceed) the final time in the second call to unique (i.e. within []).

If we were willing to do something like add a dt_primary_key class to the primary key, we could also achieve this speed in the first approach by writing a unique.dt_primary_key method, but I'm not sure how extensible this is to multiple keys (S4?)

Dominant language
R
Stars
3.9k
Forks
1.1k
Avg merge
14h 4m
Merged PRs (30d)
4

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from Rdatatable/data.table

All issues in Rdatatable/data.table

Similar issues

More R issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.