matchmaker R package

Lifecycle: experimental CRAN status Travis build status AppVeyor build status Codecov test coverage

The goal of {matchmaker} is to provide dictionary-based cleaning for R users in a simple and intuitive manner built on the {forcats} package. Some of the features of this package include:

Installation

You can install {matchmaker} from CRAN:

install.packages("matchmaker")

Example

The matchmaker package has two user-facing functions that perform dictionary-based cleaning:

Each of these functions have four manditory options:

Mostly, users will be working with match_df() to transform values across specific columns. A typical workflow would be to:

  1. construct your dictionary in a spreadsheet program based on your data
  2. read in your data and dictionary to data frames in R
  3. match!
library("matchmaker")

# Read in data set
dat <- read.csv(matchmaker_example("coded-data.csv"),
  stringsAsFactors = FALSE
)
dat$date <- as.Date(dat$date)

# Read in dictionary
dict <- read.csv(matchmaker_example("spelling-dictionary.csv"),
  stringsAsFactors = FALSE
)

Data

This is the top of our data set, generated for example purposes

id date readmission treated facility age_group lab_result_01 lab_result_02 lab_result_03 has_symptoms followup
ef267c 2019-07-08 NA 0 C 10 unk high inc NA u
e80a37 2019-07-07 y 0 3 10 inc unk norm y oui
b72883 2019-07-07 y 1 8 30 inc norm inc oui
c9ee86 2019-07-09 n 1 4 40 inc inc unk y oui
40bc7a 2019-07-12 n 1 6 0 norm unk norm NA n
46566e 2019-07-14 y NA B 50 unk unk inc NA NA

Dictionary

The dictionary looks like this:

options values grp orders
y Yes readmission 1
n No readmission 2
u Unknown readmission 3
.missing Missing readmission 4
0 Yes treated 1
1 No treated 2
.missing Missing treated 3
1 Facility 1 facility 1
2 Facility 2 facility 2
3 Facility 3 facility 3
4 Facility 4 facility 4
5 Facility 5 facility 5
6 Facility 6 facility 6
7 Facility 7 facility 7
8 Facility 8 facility 8
9 Facility 9 facility 9
10 Facility 10 facility 10
.default Unknown facility 11
0 0-9 age_group 1
10 10-19 age_group 2
20 20-29 age_group 3
30 30-39 age_group 4
40 40-49 age_group 5
50 50+ age_group 6
high High .regex ^lab_result_ 1
norm Normal .regex ^lab_result_ 2
inc Inconclusive .regex ^lab_result_ 3
y yes .global Inf
n no .global Inf
u unknown .global Inf
unk unknown .global Inf
oui yes .global Inf
.missing missing .global Inf

Matching

# Clean spelling based on dictionary -----------------------------
cleaned <- match_df(dat,
  dictionary = dict,
  from = "options",
  to = "values",
  by = "grp"
)
head(cleaned)
#>       id       date readmission treated    facility age_group
#> 1 ef267c 2019-07-08     Missing     Yes     Unknown     10-19
#> 2 e80a37 2019-07-07         Yes     Yes Facility  3     10-19
#> 3 b72883 2019-07-07         Yes      No Facility  8     30-39
#> 4 c9ee86 2019-07-09          No      No Facility  4     40-49
#> 5 40bc7a 2019-07-12          No      No Facility  6       0-9
#> 6 46566e 2019-07-14         Yes Missing     Unknown       50+
#>   lab_result_01 lab_result_02 lab_result_03 has_symptoms followup
#> 1       unknown          High  Inconclusive      missing  unknown
#> 2  Inconclusive       unknown        Normal          yes      yes
#> 3  Inconclusive        Normal  Inconclusive      missing      yes
#> 4  Inconclusive  Inconclusive       unknown          yes      yes
#> 5        Normal       unknown        Normal      missing       no
#> 6       unknown       unknown  Inconclusive      missing  missing