Rolling join does not retrieve duplicated date in x

Open
#3,180 7 comments 2 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Assessment

Difficulty
4/5
Estimated time
3-5 days
Newbie friendliness
35/100
Issue type
Bug
Clarity
Mostly clear
Activity status
Stale
Tech stack
r
Domain
data

Research direction

Reproduce the rolling and non-rolling join examples in the issue with duplicated dates and different row ordering. Read the rolling-join behavior in the main help page and determine whether the accepted resolution is a behavior change, documentation, or both. Done means the duplicate-key and ordering behavior is intentional, tested, and clearly documented.

Written by the indexing model from the issue text.

Description

joins non-equi joins

Rolling joins take into account only last occurrence of a duplicated key in x. I think this is both unexpected to the average user and inconsistent with the non-rolling join (aka indexing) which does respect duplicates. The worst part is that currently rolling joins depend on the sorting of the data.

For example

(tx <- data.table(ticker = c("BB", "AA"),
                  date = as.Date(c("1987-05-29", "1987-05-29"))))
##    ticker       date
## 1:     BB 1987-05-29
## 2:     AA 1987-05-29

(ty <- data.table(date = as.Date("1987-05-31")))
##          date
## 1: 1987-05-31

tx[ty, on = "date", roll = T]
##    ticker       date
## 1:     AA 1987-05-31

## Alternative sorting results in different join!!
setkey(tx, ticker)
tx[ty, on = "date", roll = T]
##    ticker       date
## 1:     BB 1987-05-31

## Non rolling join (subsetting) does pick the duplicates
tz <- data.table(date = as.Date("1987-05-29"))
tx[tz, on = "date"]
##    ticker       date
## 1:     AA 1987-05-29
## 2:     BB 1987-05-29

Because of this subtle issue data corruption can easily go unnoticed with large data. In so many years of using data.table this behavior was never apparent to me and it took me a while to track it down when I noticed it. So at the very least I think the missing match and order dependence must be clearly noted in the main help page.

The simplest workaround I could think of is to create a separate cross-join first and then merge on both ticker and date:

cj <- CJ(ticker = unique(tx$ticker), date = unique(ty$date))
tx[cj, on = c("ticker", "date")]
##    ticker       date
## 1:     AA 1987-05-31
## 2:     BB 1987-05-31

EDIT: The above workaround works for this simple case, but for general case where ty is multi-column I don't see a straightforward solution.

devtools::session_info()
─ Session info ─────────────────────────────────────────────────────────────────────────────────────────────────────
 setting  value                                             
 version  R Under development (unstable) (2018-12-01 r75733)
 os       Ubuntu 18.04.1 LTS                                
 system   x86_64, linux-gnu                                 
 ui       X11                                               
 language                                                   
 collate  en_GB.UTF-8                                       
 ctype    en_GB.UTF-8                                       
 tz       Europe/Amsterdam                                  
 date     2018-12-04                                        

─ Packages ─────────────────────────────────────────────────────────────────────────────────────────────────────────
 package     * version    date       lib source                                
 assertthat    0.2.0      2017-04-11 [1] CRAN (R 3.5.0)                        
 backports     1.1.2      2017-12-13 [1] CRAN (R 3.5.0)                        
 callr         3.0.0.9002 2018-12-03 [1] Github (r-lib/callr@1cbce9d)          
 cli           1.0.1      2018-09-25 [1] CRAN (R 3.6.0)                        
 crayon        1.3.4      2017-09-16 [1] CRAN (R 3.5.0)                        
 data.table  * 1.11.9     2018-12-03 [1] Github (Rdatatable/data.table@e6913a4)
 desc          1.2.0      2018-05-01 [1] CRAN (R 3.6.0)                        
 devtools      2.0.1.9000 2018-12-03 [1] Github (r-lib/devtools@a6bbe87)       
 digest        0.6.18     2018-10-10 [1] CRAN (R 3.6.0)                        
 fs            1.2.6      2018-08-23 [1] CRAN (R 3.6.0)                        
 glue          1.3.0      2018-07-17 [1] CRAN (R 3.6.0)                        
 magrittr      1.5        2014-11-22 [1] CRAN (R 3.5.0)                        
 memoise       1.1.0      2017-04-21 [1] CRAN (R 3.5.0)                        
 pkgbuild      1.0.2.9000 2018-12-03 [1] Github (r-lib/pkgbuild@6e4ebdf)       
 pkgload       1.0.1.9000 2018-12-03 [1] Github (r-lib/pkgload@0ef4f58)        
 prettyunits   1.0.2      2015-07-13 [1] CRAN (R 3.5.0)                        
 processx      3.2.0.9001 2018-12-03 [1] Github (r-lib/processx@366475b)       
 ps            1.2.1      2018-11-06 [1] CRAN (R 3.6.0)                        
 R6            2.3.0      2018-10-04 [1] CRAN (R 3.6.0)                        
 Rcpp          1.0.0      2018-11-07 [1] CRAN (R 3.6.0)                        
 remotes       2.0.2.9000 2018-12-03 [1] Github (r-lib/remotes@2e3bf35)        
 rlang         0.3.0.9000 2018-12-03 [1] Github (r-lib/rlang@035d279)          
 rprojroot     1.3-2      2018-01-03 [1] CRAN (R 3.5.0)                        
 sessioninfo   1.1.1      2018-11-05 [1] CRAN (R 3.6.0)                        
 testthat      2.0.0      2017-12-13 [1] CRAN (R 3.5.0)                        
 usethis       1.4.0      2018-08-14 [1] CRAN (R 3.6.0)                        
 withr         2.1.2.9000 2018-10-23 [1] Github (jimhester/withr@be57595)      

[1] /home/vspinu/.lib/3.6.0
[2] /home/vspinu/bin/R-latest/library
> 
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.