Data scientists, according to interviews and expert estimates, spend from 50 percent to 80 percent of their time mired in this more mundane labor of collecting and preparing unruly digital data, before it can be explored for useful nuggets.
-- "For Big-Data Scientists, 'Janitor Work' Is Key Hurdle to Insight" - The New York Times, 2014
janitor has simple functions for examining and cleaning dirty data. It was built with beginning and intermediate R users in mind and is optimized for user-friendliness. Advanced R users can already do everything covered here, but with janitor they can do it faster and save their thinking for the fun stuff.
The main janitor functions:
- perfectly format data.frame column names;
- create and format frequency tables of one, two, or three variables - think an improved
table()
; and - isolate partially-duplicate records.
The tabulate-and-report functions approximate popular features of SPSS and Microsoft Excel.
janitor is a #tidyverse-oriented package. Specifically, it plays nicely with the %>%
pipe and is optimized for cleaning data brought in with the readr and readxl packages.
You can install:
-
the most recent officially-released version from CRAN with
install.packages("janitor")
-
the latest development version from GitHub with
install.packages("devtools") devtools::install_github("sfirke/janitor")
A full description of each function, organized by topic, can be found in janitor's catalog of functions vignette. There you will find functions not mentioned in this README, like compare_df_cols()
which provides a summary of differences in column names and types when given a set of data.frames.
Below are quick examples of how janitor tools are commonly used.
Take this roster of teachers at a fictional American high school, stored in the Microsoft Excel file dirty_data.xlsx:
Dirtiness includes:
- Dreadful column names
- Rows and columns containing Excel formatting but no data
- Dates stored as numbers
- Values spread inconsistently over the "Certification" columns
Here's that data after being read in to R:
library(pacman) # for loading packages
p_load(readxl, janitor, dplyr, here)
roster_raw <- read_excel(here("dirty_data.xlsx")) # available at http://github.com/sfirke/janitor
glimpse(roster_raw)
#> Observations: 13
#> Variables: 11
#> $ `First Name` <chr> "Jason", "Jason", "Alicia", "Ada", "Desus", "Chien-Shiung", "Chien-Shiung", NA,…
#> $ `Last Name` <chr> "Bourne", "Bourne", "Keys", "Lovelace", "Nice", "Wu", "Wu", NA, "Joyce", "Lamar…
#> $ `Employee Status` <chr> "Teacher", "Teacher", "Teacher", "Teacher", "Administration", "Teacher", "Teach…
#> $ Subject <chr> "PE", "Drafting", "Music", NA, "Dean", "Physics", "Chemistry", NA, "English", "…
#> $ `Hire Date` <dbl> 39690, 39690, 37118, 27515, 41431, 11037, 11037, NA, 32994, 27919, 42221, 34700…
#> $ `% Allocated` <dbl> 0.75, 0.25, 1.00, 1.00, 1.00, 0.50, 0.50, NA, 0.50, 0.50, NA, NA, 0.80
#> $ `Full time?` <chr> "Yes", "Yes", "Yes", "Yes", "Yes", "Yes", "Yes", NA, "No", "No", "No", "No", "N…
#> $ `do not edit! --->` <lgl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA
#> $ Certification...9 <chr> "Physical ed", "Physical ed", "Instr. music", "PENDING", "PENDING", "Science 6-…
#> $ Certification...10 <chr> "Theater", "Theater", "Vocal music", "Computers", NA, "Physics", "Physics", NA,…
#> $ Certification...11 <lgl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA
Excel formatting led to an untitled empty column and 5 empty rows at the bottom of the table (only 12 records have any actual data). Bad column names are preserved.
Clean it with janitor functions:
roster <- roster_raw %>%
clean_names() %>%
remove_empty(c("rows", "cols")) %>%
mutate(hire_date = excel_numeric_to_date(hire_date),
cert = coalesce(certification_9, certification_10)) %>% # from dplyr
select(-certification_9, -certification_10) # drop unwanted columns
roster
#> # A tibble: 12 x 8
#> first_name last_name employee_status subject hire_date percent_allocated full_time cert
#> <chr> <chr> <chr> <chr> <date> <dbl> <chr> <chr>
#> 1 Jason Bourne Teacher PE 2008-08-30 0.75 Yes Physical ed
#> 2 Jason Bourne Teacher Drafting 2008-08-30 0.25 Yes Physical ed
#> 3 Alicia Keys Teacher Music 2001-08-15 1 Yes Instr. music
#> 4 Ada Lovelace Teacher <NA> 1975-05-01 1 Yes PENDING
#> 5 Desus Nice Administration Dean 2013-06-06 1 Yes PENDING
#> 6 Chien-Shiung Wu Teacher Physics 1930-03-20 0.5 Yes Science 6-12
#> 7 Chien-Shiung Wu Teacher Chemistry 1930-03-20 0.5 Yes Science 6-12
#> 8 James Joyce Teacher English 1990-05-01 0.5 No English 6-12
#> 9 Hedy Lamarr Teacher Science 1976-06-08 0.5 No PENDING
#> 10 Carlos Boozer Coach Basketball 2015-08-05 NA No Physical ed
#> 11 Young Boozer Coach <NA> 1995-01-01 NA No Political sci.
#> 12 Micheal Larsen Teacher English 2009-09-15 0.8 No Vocal music
Name cleaning comes in two flavors, the main make_clean_names()
and the convenience version clean_names()
for piped data.frame workflows. The core janitor cleaning function is clean_names()
- call it whenever you load data into R as in the example above.
make_clean_names()
operates on character vectors and can be used during data import. Here's an example of how it can be used:
library(pacman) # for loading packages
p_load(readxl, janitor, dplyr, here)
roster_raw <- read_excel(here("dirty_data.xlsx"), .name_repair = make_clean_names) # available at http://github.com/sfirke/janitor
glimpse(roster_raw)
#> Observations: 13
#> Variables: 11
#> $ first_name <chr> "Jason", "Jason", "Alicia", "Ada", "Desus", "Chien-Shiung", "Chien-Shiung", NA, "…
#> $ last_name <chr> "Bourne", "Bourne", "Keys", "Lovelace", "Nice", "Wu", "Wu", NA, "Joyce", "Lamarr"…
#> $ employee_status <chr> "Teacher", "Teacher", "Teacher", "Teacher", "Administration", "Teacher", "Teacher…
#> $ subject <chr> "PE", "Drafting", "Music", NA, "Dean", "Physics", "Chemistry", NA, "English", "Sc…
#> $ hire_date <dbl> 39690, 39690, 37118, 27515, 41431, 11037, 11037, NA, 32994, 27919, 42221, 34700, …
#> $ percent_allocated <dbl> 0.75, 0.25, 1.00, 1.00, 1.00, 0.50, 0.50, NA, 0.50, 0.50, NA, NA, 0.80
#> $ full_time <chr> "Yes", "Yes", "Yes", "Yes", "Yes", "Yes", "Yes", NA, "No", "No", "No", "No", "No"
#> $ do_not_edit <lgl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA
#> $ certification <chr> "Physical ed", "Physical ed", "Instr. music", "PENDING", "PENDING", "Science 6-12…
#> $ certification_2 <chr> "Theater", "Theater", "Vocal music", "Computers", NA, "Physics", "Physics", NA, "…
#> $ certification_3 <lgl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA
Use get_dupes()
to identify and examine duplicate records during data cleaning. Let's see if any teachers are listed more than once:
roster %>% get_dupes(first_name, last_name)
#> # A tibble: 4 x 9
#> first_name last_name dupe_count employee_status subject hire_date percent_allocat… full_time cert
#> <chr> <chr> <int> <chr> <chr> <date> <dbl> <chr> <chr>
#> 1 Chien-Shiung Wu 2 Teacher Physics 1930-03-20 0.5 Yes Science 6…
#> 2 Chien-Shiung Wu 2 Teacher Chemistry 1930-03-20 0.5 Yes Science 6…
#> 3 Jason Bourne 2 Teacher PE 2008-08-30 0.75 Yes Physical …
#> 4 Jason Bourne 2 Teacher Drafting 2008-08-30 0.25 Yes Physical …
Yes, some teachers appear twice. We ought to address this before counting employees.
A variable (or combinations of two or three variables) can be tabulated with tabyl()
. The resulting data.frame can be tweaked and formatted with the suite of adorn_
functions for quick analysis and printing of pretty results in a report. adorn_
functions can be helpful with non-tabyls, too.
tabyl
can be called two ways:
- On a vector, when tabulating a single variable - e.g.,
tabyl(roster$subject)
- On a data.frame, specifying 1, 2, or 3 variable names to tabulate :
roster %>% tabyl(subject, employee_status)
.- Here the data.frame is passed in with the
%>%
pipe; this allowstabyl
to be used in an analysis pipeline
- Here the data.frame is passed in with the
Like table()
, but pipe-able, data.frame-based, and fully featured.
One variable:
roster %>%
tabyl(subject)
#> subject n percent valid_percent
#> Basketball 1 0.08333333 0.1
#> Chemistry 1 0.08333333 0.1
#> Dean 1 0.08333333 0.1
#> Drafting 1 0.08333333 0.1
#> English 2 0.16666667 0.2
#> Music 1 0.08333333 0.1
#> PE 1 0.08333333 0.1
#> Physics 1 0.08333333 0.1
#> Science 1 0.08333333 0.1
#> <NA> 2 0.16666667 NA
Two variables:
roster %>%
filter(hire_date > as.Date("1950-01-01")) %>%
tabyl(employee_status, full_time)
#> employee_status No Yes
#> Administration 0 1
#> Coach 2 0
#> Teacher 3 4
Three variables:
roster %>%
tabyl(full_time, subject, employee_status, show_missing_levels = FALSE)
#> $Administration
#> full_time Dean
#> Yes 1
#>
#> $Coach
#> full_time Basketball NA_
#> No 1 1
#>
#> $Teacher
#> full_time Chemistry Drafting English Music PE Physics Science NA_
#> No 0 0 2 0 0 0 1 0
#> Yes 1 1 0 1 1 1 0 1
The adorn_
functions dress up the results of these tabulation calls for fast, basic reporting. Here are some of the functions that augment a summary table for reporting:
roster %>%
tabyl(employee_status, full_time) %>%
adorn_totals("row") %>%
adorn_percentages("row") %>%
adorn_pct_formatting() %>%
adorn_ns() %>%
adorn_title("combined")
#> employee_status/full_time No Yes
#> Administration 0.0% (0) 100.0% (1)
#> Coach 100.0% (2) 0.0% (0)
#> Teacher 33.3% (3) 66.7% (6)
#> Total 41.7% (5) 58.3% (7)
Pipe that right into knitr::kable()
in your RMarkdown report.
These modular adornments can be layered to reduce R's deficit against Excel and SPSS when it comes to quick, informative counts.
You are welcome to:
- submit suggestions and report bugs: https://github.com/sfirke/janitor/issues
- let me know what you think on twitter @samfirke
- compose a friendly e-mail to: