statamerge

statamerge is a dplyr wrapper designed to add the functionality from statamerges to dplyr merges. Specifically, it checks for uniqueness, adds a source “merge code” to show whether an observation was matched or not, and coalesces variables from the master and using with the same name (keeping master if there is a conflict).

Installation

You can install the development version of statamerge from GitHub with:

# install.packages("devtools")
devtools::install_github("janetmalzahn/statamerge")

Why use statamerge?

Statamerge is particularly useful for those learning R with a background in Stata. However, Stata’s merges have plenty of useful features for those frequently working with merged datasets:

1. statamerge checks for uniqueness to ensure the dataset matches expectations. A 1:1 merge checks for uniqueness in each dataset to ensure that each observation in the master can will only be matched to a (maximum) of one observation in using. A 1:m or m:1 checks for uniqueness in either the master (m:1) or the using (1:m) datasets. This ensures that the merged dataset will have no more observations than the non-unique dataset.

2. statamerge includes a code for whether an observation was matched, unmatched from the master dataset, or unmatched from the using. With current dplyr merges, there is no simple way to distinguish items that were matched from items that were not, or which dataset unamtched observations came from. statamerge outputs a dataset with a merge code that is 1 if the observation was unmatched from the master, 2 if the observation was unmatched from using, or 3 if the observation was merged.

3. statamerge coalesces variables that share the same name in master and using into one dataset. Instead of creating two separate variables with a suffix indicating which dataset the observations came from, it combines one into one column.

Example

statamerge uses language from stata to refer to datasets. The first dataset input to statamerge is the “master” dataset and the second is the “using” dataset.

library(statamerge)
# make datasets
merge1 <- dplyr::tibble(band = c("Big Thief", "Los Campesinos!", "Spoon", "Remi Wolf"),
                       type = c("band", "band", "band", "singer"),
                       genre = c("folk", "twee", "indie", "funk"))
merge2 <- dplyr::tibble(song = c("Songs About Your Girlfriend", "Do You", "Michael", "Jesus Etc"), band = c("Los Campesinos!", "Spoon", "Remi Wolf", "Wilco"), type = c("banger", "jam", "banger", "jam"))

merge3 <- dplyr::tibble(song = c("Songs About Your Girlfriend", "Do You", "Michael", "Jesus Etc", "The Underdog"),
                       band = c("Los Campesinos!", "Spoon", "Remi Wolf", "Wilco", "Spoon"),
                       type = c("banger", "jam", "banger", "jam", "banger"))

# one to one merge
statamerge(merge1, merge2, mergetype = "1:1", merge_vars = "band")
#> [1] "Performing 1:1 merge"
#> # A tibble: 3 × 2
#>   merge_code     n
#>        <dbl> <int>
#> 1          1     1
#> 2          2     1
#> 3          3     3
#> # A tibble: 5 × 5
#>   band            genre song                        type   merge_code
#>   <chr>           <chr> <chr>                       <chr>       <dbl>
#> 1 Big Thief       folk  <NA>                        band            1
#> 2 Los Campesinos! twee  Songs About Your Girlfriend band            3
#> 3 Spoon           indie Do You                      band            3
#> 4 Remi Wolf       funk  Michael                     singer          3
#> 5 Wilco           <NA>  Jesus Etc                   jam             2

# stata merge 1:m
statamerge(merge1, merge3, mergetype = "1:m", merge_vars = "band")
#> [1] "Performing 1:m merge"
#> # A tibble: 3 × 2
#>   merge_code     n
#>        <dbl> <int>
#> 1          1     1
#> 2          2     1
#> 3          3     4
#> # A tibble: 6 × 5
#>   band            genre song                        type   merge_code
#>   <chr>           <chr> <chr>                       <chr>       <dbl>
#> 1 Big Thief       folk  <NA>                        band            1
#> 2 Los Campesinos! twee  Songs About Your Girlfriend band            3
#> 3 Spoon           indie Do You                      band            3
#> 4 Spoon           indie The Underdog                band            3
#> 5 Remi Wolf       funk  Michael                     singer          3
#> 6 Wilco           <NA>  Jesus Etc                   jam             2

# stata merge 1:m
statamerge(merge3, merge1, mergetype = "m:1", merge_vars = "band")
#> [1] "Performing m:1 merge"
#> # A tibble: 3 × 2
#>   merge_code     n
#>        <dbl> <int>
#> 1          1     1
#> 2          2     1
#> 3          3     4
#> # A tibble: 6 × 5
#>   song                        band            genre type   merge_code
#>   <chr>                       <chr>           <chr> <chr>       <dbl>
#> 1 Songs About Your Girlfriend Los Campesinos! twee  banger          3
#> 2 Do You                      Spoon           indie jam             3
#> 3 Michael                     Remi Wolf       funk  banger          3
#> 4 Jesus Etc                   Wilco           <NA>  jam             1
#> 5 The Underdog                Spoon           indie banger          3
#> 6 <NA>                        Big Thief       folk  band            2