Working with Strings in R
1 Introduction
Proficiency in string manipulation is a vital skill for data scientists. Tasks like cleaning messy data and formatting outputs rely heavily on the ability to parse, combine, and modify character strings. This lesson focuses on techniques for working with strings in R, utilizing functions from the {stringr} package in the tidyverse. Let’s dive in!
2 Learning Objectives
- Understand the concept of strings and rules for defining them in R
- Use escapes to include special characters like quotes within strings
- Employ {stringr} functions to format strings:
- Change case with
str_to_lower(),str_to_upper(),str_to_title() - Trim whitespace with
str_trim()andstr_squish() - Pad strings to equal width with
str_pad() - Wrap text to a certain width using
str_wrap()
- Change case with
- Split strings into parts using
str_split()andseparate() - Combine strings together with
paste()andpaste0() - Extract substrings from strings using
str_sub()
3 Packages
4 Defining Strings
There are fundamental rules for defining character strings in R.
Strings can be enclosed in either single or double quotes. However, the type of quotation mark used at the start must match the one used at the end. For example:
You cannot normally include double quotes inside a string that starts and ends with double quotes. The same applies to single quotes inside a string that starts and ends with single quotes. For example:
will_not_work <- "Double quotes " inside double quotes"
will_not_work <- 'Single quotes ' inside double quotes'But you can include single quotes inside a string that starts and ends with double quotes, and vice versa:
Alternatively, you can use the escape character \ to
include a literal single or double quote inside a string:
single_quote <- 'Single quotes \' inside double quotes'
double_quote <- "Double quotes \" inside double quotes"To display these strings as they would appear in output, such as on a
plot, use cat():
## Single quotes ' inside double quotes## Double quotes " inside double quotescat() prints its arguments without additional
formatting.
Since \ is the escape character, you must use
\\ to include a literal backslash in a string:
## This is a backslash: \Q: Error Spotting in String Definitions
Below are attempts to define character strings in R, with two out of five lines containing an error. Identify and correct these errors.
5 String Formatting in R with {stringr}
The {stringr} package in R provides useful functions for formatting strings for analysis and visualization. This includes case changes, whitespace handling, length standardization, and text wrapping.
5.1 Changing Case
Converting case is often needed to standardize strings or prepare them for display. The {stringr} package provides several case-changing functions:
str_to_upper()converts strings to uppercase.
## [1] "HELLO WORLD"str_to_lower()converts strings to lowercase.
## [1] "goodbye"str_to_title()capitalizes the first letter of each word. Ideal for titling names, subjects, etc.
## [1] "String Manipulation"5.2 Handling Whitespace
Managing whitespace makes strings neat and uniform. The {stringr} package provides two main functions for this:
str_trim()removes whitespace at the start and end.
## [1] "trimmed"str_squish()removes whitespace at the start and end, and reduces multiple internal spaces to one.
## [1] "too much space"## [1] "too much space"5.3 Text Padding
str_pad() pads strings to a fixed width. For example, we
can pad the number 7 to force it to have 3 characters:
## [1] "007"The first argument is the string to pad. width sets the
final string width and pad specifies the padding
character.
side controls whether padding is added on the left or
right. The side argument defaults to “left”, so padding
will be added on the left side if not specified. Specifying
side = "right" pads on the right side instead:
## [1] "7___"Or we can pad on both sides:
## [1] "__7__"5.4 Text Wrapping
Text wrapping helps fit strings into confined spaces like plot
titles. The str_wrap() function wraps text to a set
width.
For example, to wrap text at 10 characters we can write:
example_string <- "String Manipulation with str_wrap can enhance readability in plots."
wrapped_to_10 <- str_wrap(example_string, width = 10)
wrapped_to_10## [1] "String\nManipulation\nwith\nstr_wrap\ncan\nenhance\nreadability\nin plots."The output may appear confusing. The \n indicates a line
break, and to view the modified properly, we need to use the
cat() function, which is a special version of
print():
## String
## Manipulation
## with
## str_wrap
## can
## enhance
## readability
## in plots.Notice that the function maintains whole words, so it won’t split longer words like “manipulation”.
Setting the width to 1 essentially splits the string into individual words:
## String
## Manipulation
## with
## str_wrap
## can
## enhance
## readability
## in
## plots.str_wrap() is particularly useful in plotting with
ggplot2. For example, wrapping a long title to prevent it from spilling
over the plot:
long_title <- "This is an example of a very long title, which would usually run over the end of your ggplot, but you can wrap it with str_wrap to fit within a specified character limit."
# Without wrapping
ggplot(women, aes(height, weight)) +
geom_point() +
labs(title = long_title)
# With wrapping at 80 characters
ggplot(women, aes(height, weight)) +
geom_point() +
labs(title = str_wrap(long_title, width = 50))
So str_wrap() keeps titles neatly within the plot!
Q: Cleaning Patient Name Data
A dataset contains patient names with inconsistent formatting and extra white spaces. Use the {stringr} package to standardize this information:
Q: Standardizing Drug Codes
The following (fictional) drug codes are inconsistently formatted. Standardize them by padding with zeros to ensure all codes are 8 characters long:
Q: Wrapping Medical Instructions
Use str_wrap() to format the following for better
readability:
instructions <- "Take two tablets daily after meals. If symptoms persist for more than three days, consult your doctor immediately. Do not take more than the recommended dose. Keep out of reach of children."
ggplot(data.frame(x = 1, y = 1), aes(x, y, label = instructions)) +
geom_label() +
theme_void()
5.5 Applying String Formatting to a Dataset
Now let’s apply the {stringr} package’s string formatting functions to clean and standardize a dataset. Our focus is on a dataset from a study on HIV care and treatment services in Zambézia Province, Mozambique, available here. The original dataset had various formatting inconsistencies, but we’ve added additional mistakes for educational purposes.
First, we load the dataset and examine specific variables for potential issues.
# Load the dataset
hiv_dat_messy_1 <- openxlsx::read.xlsx(here("data/hiv_dat_messy_1.xlsx")) %>%
as_tibble()
# These four variables contain formatting inconsistencies:
hiv_dat_messy_1 %>%
select(district, health_unit, education, regimen)## # A tibble: 1,413 × 4
## district health_unit education regimen
## <chr> <chr> <chr> <chr>
## 1 "Rural" District Hospital Maganj… MISSING AZT+3TC+NVP
## 2 "Rural" District Hospital Maganj… secondary TDF+3TC+EFV
## 3 "Urban" 24th Of July Health … MISSING tdf+3tc+efv
## 4 "Urban" 24th Of July Health … MISSING TDF+3TC+EFV
## 5 " Urban" 24th Of July Health … University tdf+3tc+efv
## 6 "Urban" 24th Of July Health Faci… Technical AZT+3TC+NVP
## 7 "Rural" District Hospital Maganj… Technical TDF+3TC+EFV
## 8 "Urban" 24th Of July Health Faci… Technical azt+3tc+nvp
## 9 "Urban" 24th Of July Health Faci… Technical AZT+3TC+NVP
## 10 "Urban" 24th Of July Health Faci… Technical TDF+3TC+EFV
## # ℹ 1,403 more rowsUsing the tabyl function, we can identify and count
unique values, revealing the inconsistencies:
## health_unit n percent
## 24th Of July Health Facility 239 0.16914367
## 24th Of July Health Facility 249 0.17622081
## District Hospital Maganja Da Costa 342 0.24203822
## District Hospital Maganja Da Costa 336 0.23779193
## Nante Health Facility 119 0.08421798
## Nante Health Facility 128 0.09058740## education n percent
## MISSING 776 0.549186129
## None 128 0.090587403
## Primary 178 0.125973107
## Secondary 82 0.058032555
## Technical 17 0.012031139
## University 4 0.002830856
## primary 157 0.111111111
## secondary 71 0.050247700## regimen n percent valid_percent
## AZT+3TC+EFV 24 0.0169851380 0.0179910045
## AZT+3TC+NVP 229 0.1620665251 0.1716641679
## D4T+3TC+ABC 1 0.0007077141 0.0007496252
## D4T+3TC+EFV 2 0.0014154282 0.0014992504
## D4T+3TC+NVP 16 0.0113234253 0.0119940030
## OTHER 1 0.0007077141 0.0007496252
## TDF+3TC+EFV 404 0.2859164897 0.3028485757
## TDF+3TC+NVP 3 0.0021231423 0.0022488756
## azt+3tc+efv 16 0.0113234253 0.0119940030
## azt+3tc+nvp 231 0.1634819533 0.1731634183
## d4t+3tc+efv 9 0.0063694268 0.0067466267
## d4t+3tc+nvp 18 0.0127388535 0.0134932534
## d4t+4tc+nvp 1 0.0007077141 0.0007496252
## d4t6+3tc+nvp 2 0.0014154282 0.0014992504
## other 2 0.0014154282 0.0014992504
## tdf+3tc+efv 374 0.2646850672 0.2803598201
## tdf+3tc+nvp 1 0.0007077141 0.0007496252
## <NA> 79 0.0559094126 NA## district n percent
## Rural 234 0.16560510
## Urban 118 0.08351026
## Rural 691 0.48903043
## Urban 370 0.26185421Another useful function for visualizing these issues is
tbl_summary from the {gtsummary} package:
| Characteristic | N = 1,4131 |
|---|---|
| district | |
| Rural | 234 (17%) |
| Urban | 118 (8.4%) |
| Rural | 691 (49%) |
| Urban | 370 (26%) |
| health_unit | |
| 24th Of July Health Facility | 239 (17%) |
| 24th Of July Health Facility | 249 (18%) |
| District Hospital Maganja Da Costa | 342 (24%) |
| District Hospital Maganja Da Costa | 336 (24%) |
| Nante Health Facility | 119 (8.4%) |
| Nante Health Facility | 128 (9.1%) |
| education | |
| MISSING | 776 (55%) |
| None | 128 (9.1%) |
| primary | 157 (11%) |
| Primary | 178 (13%) |
| secondary | 71 (5.0%) |
| Secondary | 82 (5.8%) |
| Technical | 17 (1.2%) |
| University | 4 (0.3%) |
| regimen | |
| azt+3tc+efv | 16 (1.2%) |
| AZT+3TC+EFV | 24 (1.8%) |
| azt+3tc+nvp | 231 (17%) |
| AZT+3TC+NVP | 229 (17%) |
| D4T+3TC+ABC | 1 (<0.1%) |
| d4t+3tc+efv | 9 (0.7%) |
| D4T+3TC+EFV | 2 (0.1%) |
| d4t+3tc+nvp | 18 (1.3%) |
| D4T+3TC+NVP | 16 (1.2%) |
| d4t+4tc+nvp | 1 (<0.1%) |
| d4t6+3tc+nvp | 2 (0.1%) |
| other | 2 (0.1%) |
| OTHER | 1 (<0.1%) |
| tdf+3tc+efv | 374 (28%) |
| TDF+3TC+EFV | 404 (30%) |
| tdf+3tc+nvp | 1 (<0.1%) |
| TDF+3TC+NVP | 3 (0.2%) |
| Unknown | 79 |
| 1 n (%) | |
The output clearly shows inconsistencies in casing, spacing, and format, so we need to standardize them.
Next, we address these issues systematically:
hiv_dat_clean_1 <- hiv_dat_messy_1 %>%
mutate(
district = str_to_title(str_trim(district)), # Standardizing district names
health_unit = str_squish(health_unit), # Removing extra spaces
education = str_to_title(education), # Standardizing education levels
regimen = str_to_upper(regimen) # Consistency in regimen column
)And we can verify the effectiveness of these changes by rerunning the
tbl_summary() function:
| Characteristic | N = 1,4131 |
|---|---|
| district | |
| Rural | 925 (65%) |
| Urban | 488 (35%) |
| health_unit | |
| 24th Of July Health Facility | 488 (35%) |
| District Hospital Maganja Da Costa | 678 (48%) |
| Nante Health Facility | 247 (17%) |
| education | |
| Missing | 776 (55%) |
| None | 128 (9.1%) |
| Primary | 335 (24%) |
| Secondary | 153 (11%) |
| Technical | 17 (1.2%) |
| University | 4 (0.3%) |
| regimen | |
| AZT+3TC+EFV | 40 (3.0%) |
| AZT+3TC+NVP | 460 (34%) |
| D4T+3TC+ABC | 1 (<0.1%) |
| D4T+3TC+EFV | 11 (0.8%) |
| D4T+3TC+NVP | 34 (2.5%) |
| D4T+4TC+NVP | 1 (<0.1%) |
| D4T6+3TC+NVP | 2 (0.1%) |
| OTHER | 3 (0.2%) |
| TDF+3TC+EFV | 778 (58%) |
| TDF+3TC+NVP | 4 (0.3%) |
| Unknown | 79 |
| 1 n (%) | |
Great!
Finally, let’s attempt to plot counts of the health_unit
variable. For the plot style below, we encounter an issue with lengthy
labels:
To resolve this, we can adjust the labels using
str_wrap():
Much cleaner, though we should probably fix the axis title:
hiv_dat_clean_1 %>%
ggplot(aes(x = str_wrap(health_unit, width = 20))) +
geom_bar() +
labs(x = "Health Unit")
Now try your hand on similar cleaning operations in the practice questions below.
Q: Formatting a Tuberculosis Dataset
In this exercise, you will clean a dataset, lima_messy,
originating from a tuberculosis (TB) treatment adherence study in Lima,
Peru. More details about the study and the dataset are available here.
Begin by importing the dataset:
## # A tibble: 1,293 × 18
## id age sex marital_status
## <chr> <chr> <chr> <chr>
## 1 pe-1008 38 and older M Single
## 2 lm-1009 38 and older M Married / cohabitating
## 3 pe-1010 27 to 37 m Married / cohabitating
## 4 lm-1011 27 to 37 m Married / cohabitating
## 5 pe-1012 38 and older m Married / cohabitating
## 6 lm-1013 27 to 37 M Single
## 7 pe-1014 27 To 37 m Married / cohabitating
## 8 lm-1015 22 To 26 m Single
## 9 pe-1016 27 to 37 m Single
## 10 lm-1017 22 to 26 m Single
## # ℹ 1,283 more rows
## # ℹ 14 more variables: poverty_level <chr>, …Your task is to clean the marital_status,
sex, and age variables in
lima_messy. Following the cleaning process, generate a
summary table using the tbl_summary() function. Aim for
your output to align with this structure:
| Characteristic | N = 1,293 |
|---|---|
| marital_status | |
| Divorced / Separated | 93 (7.2%) |
| Married / Cohabitating | 486 (38%) |
| Single | 677 (52%) |
| Widowed | 37 (2.9%) |
| sex | |
| F | 503 (39%) |
| M | 790 (61%) |
| age | |
| 21 and younger | 338 (26%) |
| 22 to 26 | 345 (27%) |
| 27 to 37 | 303 (23%) |
| 38 and older | 307 (24%) |
Implement the cleaning and summarize:
Q: Wrapping Axis Labels in a Plot
Using the cleaned dataset lima_clean from the previous
task, create a bar plot to display the count of participants by
marital_status. Then wrap the axis labels on the x-axis to
a maximum of 15 characters per line for readability.
6 Splitting Strings with
str_split() and separate()
Splitting strings is common task in data manipulation. The tidyverse
offers efficient functions for this task, notably
stringr::str_split() and
tidyr::separate().
6.1 Using
str_split()
The str_split() function is useful for dividing strings
into parts. For example:
## [[1]]
## [1] "split" "this" "string"This code splits example_string at each hyphen.
However, applying str_split() directly to a dataframe
can be more complex.
Let’s try it with the IRS dataset from Malawi as a case study. You
should already be familiar with this dataset from a previous lesson. It
is available here. For now, we’ll focus on the
start_date_long column:
irs <- read_csv(here("data/Illovo_data.csv"))
irs_dates_1 <- irs %>% select(village, start_date_long)
irs_dates_1## # A tibble: 112 × 2
## village start_date_long
## <chr> <chr>
## 1 Mess April 07 2014
## 2 Nkombedzi April 22 2014
## 3 B Compound May 13 2014
## 4 D Compound May 13 2014
## 5 Post Office May 13 2014
## 6 Mangulenje May 15 2014
## 7 Mangulenje Senior May 27 2014
## 8 Old School May 27 2014
## 9 Mwanza May 28 2014
## 10 Alumenda June 18 2014
## # ℹ 102 more rowsSuppose we want to split the start_date_long variable to
extract the day, month, and year. We can write:
## # A tibble: 112 × 3
## village start_date_long start_date_parts
## <chr> <chr> <list>
## 1 Mess April 07 2014 <chr [3]>
## 2 Nkombedzi April 22 2014 <chr [3]>
## 3 B Compound May 13 2014 <chr [3]>
## 4 D Compound May 13 2014 <chr [3]>
## 5 Post Office May 13 2014 <chr [3]>
## 6 Mangulenje May 15 2014 <chr [3]>
## 7 Mangulenje Senior May 27 2014 <chr [3]>
## 8 Old School May 27 2014 <chr [3]>
## 9 Mwanza May 28 2014 <chr [3]>
## 10 Alumenda June 18 2014 <chr [3]>
## # ℹ 102 more rowsThis results in a list column, which can be difficult to work with.
To make it more readable, we can use unnest_wider():
irs_dates_1 %>%
mutate(start_date_parts = str_split(start_date_long, " ")) %>%
unnest_wider(start_date_parts, names_sep = "_")## # A tibble: 112 × 5
## village start_date_long start_date_parts_1
## <chr> <chr> <chr>
## 1 Mess April 07 2014 April
## 2 Nkombedzi April 22 2014 April
## 3 B Compound May 13 2014 May
## 4 D Compound May 13 2014 May
## 5 Post Office May 13 2014 May
## 6 Mangulenje May 15 2014 May
## 7 Mangulenje Senior May 27 2014 May
## 8 Old School May 27 2014 May
## 9 Mwanza May 28 2014 May
## 10 Alumenda June 18 2014 June
## # ℹ 102 more rows
## # ℹ 2 more variables: start_date_parts_2 <chr>, …It works! Our date parts are now split. However, this approach is
quite cumbersome. A better solution for splitting components is the
separate() function.
6.2 Using
separate()
Let’s try the same task using separate():
## # A tibble: 112 × 4
## village month day year
## <chr> <chr> <chr> <chr>
## 1 Mess April 07 2014
## 2 Nkombedzi April 22 2014
## 3 B Compound May 13 2014
## 4 D Compound May 13 2014
## 5 Post Office May 13 2014
## 6 Mangulenje May 15 2014
## 7 Mangulenje Senior May 27 2014
## 8 Old School May 27 2014
## 9 Mwanza May 28 2014
## 10 Alumenda June 18 2014
## # ℹ 102 more rowsMuch more straightforward!
This function requires specifying:
- The column to be split.
into- Names of the new columns.sep- The separator character.
To retain the original column, use remove = FALSE:
irs_dates_1 %>%
separate(start_date_long, into = c("month", "day", "year"), sep = " ", remove = FALSE)## # A tibble: 112 × 5
## village start_date_long month day year
## <chr> <chr> <chr> <chr> <chr>
## 1 Mess April 07 2014 April 07 2014
## 2 Nkombedzi April 22 2014 April 22 2014
## 3 B Compound May 13 2014 May 13 2014
## 4 D Compound May 13 2014 May 13 2014
## 5 Post Office May 13 2014 May 13 2014
## 6 Mangulenje May 15 2014 May 15 2014
## 7 Mangulenje Senior May 27 2014 May 27 2014
## 8 Old School May 27 2014 May 27 2014
## 9 Mwanza May 28 2014 May 28 2014
## 10 Alumenda June 18 2014 June 18 2014
## # ℹ 102 more rowsAlternatively, the lubridate package offers functions to
extract date components:
irs_dates_1 %>%
mutate(start_date_long = mdy(start_date_long)) %>%
mutate(day = day(start_date_long),
month = month(start_date_long, label = TRUE),
year = year(start_date_long))## # A tibble: 112 × 5
## village start_date_long day month year
## <chr> <date> <int> <ord> <dbl>
## 1 Mess 2014-04-07 7 Apr 2014
## 2 Nkombedzi 2014-04-22 22 Apr 2014
## 3 B Compound 2014-05-13 13 May 2014
## 4 D Compound 2014-05-13 13 May 2014
## 5 Post Office 2014-05-13 13 May 2014
## 6 Mangulenje 2014-05-15 15 May 2014
## 7 Mangulenje Senior 2014-05-27 27 May 2014
## 8 Old School 2014-05-27 27 May 2014
## 9 Mwanza 2014-05-28 28 May 2014
## 10 Alumenda 2014-06-18 18 Jun 2014
## # ℹ 102 more rowsWhen some rows lack all the necessary parts, separate()
will issue a warning. Let’s demonstrate this by artificially removing
all instances of the word “April” from our dates:
irs_dates_with_problem <-
irs_dates_1 %>%
mutate(start_date_missing = str_replace(start_date_long, "April ", ""))
irs_dates_with_problem## # A tibble: 112 × 3
## village start_date_long start_date_missing
## <chr> <chr> <chr>
## 1 Mess April 07 2014 07 2014
## 2 Nkombedzi April 22 2014 22 2014
## 3 B Compound May 13 2014 May 13 2014
## 4 D Compound May 13 2014 May 13 2014
## 5 Post Office May 13 2014 May 13 2014
## 6 Mangulenje May 15 2014 May 15 2014
## 7 Mangulenje Senior May 27 2014 May 27 2014
## 8 Old School May 27 2014 May 27 2014
## 9 Mwanza May 28 2014 May 28 2014
## 10 Alumenda June 18 2014 June 18 2014
## # ℹ 102 more rowsNow, let’s try to split the date parts:
irs_dates_with_problem %>%
separate(start_date_missing, into = c("month", "day", "year"), sep = " ")## Warning: Expected 3 pieces. Missing pieces filled with `NA` in 3 rows [1, 2,
## 12].## # A tibble: 112 × 5
## village start_date_long month day year
## <chr> <chr> <chr> <chr> <chr>
## 1 Mess April 07 2014 07 2014 <NA>
## 2 Nkombedzi April 22 2014 22 2014 <NA>
## 3 B Compound May 13 2014 May 13 2014
## 4 D Compound May 13 2014 May 13 2014
## 5 Post Office May 13 2014 May 13 2014
## 6 Mangulenje May 15 2014 May 15 2014
## 7 Mangulenje Senior May 27 2014 May 27 2014
## 8 Old School May 27 2014 May 27 2014
## 9 Mwanza May 28 2014 May 28 2014
## 10 Alumenda June 18 2014 June 18 2014
## # ℹ 102 more rowsAs you can see, rows missing parts will produce warnings. Handle such warnings carefully, as they can lead to inaccurate data. In this case, we now have the day and month information for those rows in the wrong columns.
Q: Splitting Age Range Strings
Consider the esoph_ca dataset, from the {medicaldata}
package, which involves a case-control study of esophageal cancer in
France.
## # A tibble: 88 × 5
## agegp alcgp tobgp ncases ncontrols
## <ord> <ord> <ord> <dbl> <dbl>
## 1 25-34 0-39g/day 0-9g/day 0 40
## 2 25-34 0-39g/day 10-19 0 10
## 3 25-34 0-39g/day 20-29 0 6
## 4 25-34 0-39g/day 30+ 0 5
## 5 25-34 40-79 0-9g/day 0 27
## 6 25-34 40-79 10-19 0 7
## 7 25-34 40-79 20-29 0 4
## 8 25-34 40-79 30+ 0 7
## 9 25-34 80-119 0-9g/day 0 2
## 10 25-34 80-119 10-19 0 1
## # ℹ 78 more rowsSplit the age ranges in the agegp column into two
separate columns: agegp_lower and
agegp_upper.
After using the separate() function, the “75+” age group
will require special handling. Use readr::parse_number()or
another method to convert the lower age limit (“75+”) to a number.
6.3 Separating Special Characters
To use the separate() function on special characters
like the period (.), we need to escape them with a double backslash
(\\).
Consider the scenario where dates are formatted with periods:
irs_with_period <- irs_dates_1 %>%
mutate(start_date_long = format(lubridate::mdy(start_date_long), "%d.%m.%Y"))
irs_with_period## # A tibble: 112 × 2
## village start_date_long
## <chr> <chr>
## 1 Mess 07.04.2014
## 2 Nkombedzi 22.04.2014
## 3 B Compound 13.05.2014
## 4 D Compound 13.05.2014
## 5 Post Office 13.05.2014
## 6 Mangulenje 15.05.2014
## 7 Mangulenje Senior 27.05.2014
## 8 Old School 27.05.2014
## 9 Mwanza 28.05.2014
## 10 Alumenda 18.06.2014
## # ℹ 102 more rowsAttempting to separate this date format directly with
sep = "." will not work:
## # A tibble: 112 × 4
## village day month year
## <chr> <chr> <chr> <chr>
## 1 Mess "" "" ""
## 2 Nkombedzi "" "" ""
## 3 B Compound "" "" ""
## 4 D Compound "" "" ""
## 5 Post Office "" "" ""
## 6 Mangulenje "" "" ""
## 7 Mangulenje Senior "" "" ""
## 8 Old School "" "" ""
## 9 Mwanza "" "" ""
## 10 Alumenda "" "" ""
## # ℹ 102 more rowsThis doesn’t work as intended because, in regex (regular expressions), the period is a special character. We’ll learn more about these in due course. The correct approach is to escape the period uses a double backslash (\):
## # A tibble: 112 × 4
## village day month year
## <chr> <chr> <chr> <chr>
## 1 Mess 07 04 2014
## 2 Nkombedzi 22 04 2014
## 3 B Compound 13 05 2014
## 4 D Compound 13 05 2014
## 5 Post Office 13 05 2014
## 6 Mangulenje 15 05 2014
## 7 Mangulenje Senior 27 05 2014
## 8 Old School 27 05 2014
## 9 Mwanza 28 05 2014
## 10 Alumenda 18 06 2014
## # ℹ 102 more rowsNow, the function understands to split the string at each literal period.
Similarly, when using other special characters like +,
*, or ?, we also need to precede them with a
double backslash (\) in the sep argument.
What is a Special Character?
In regular expressions, which help find patterns in text, special
characters have specific roles. For example, a period (.) is a wildcard
that can represent any character. So, in a search, “do.t” could match
“dolt,” “dost,” or “doct” Similarly, the plus sign (+) is used to
indicate one or more occurrences of the preceding character. For
example, “ho+se” would match “hose” or “hooose” but not “hse.” When we
need to use these characters in their ordinary roles, we use a double
backslash (\\) before them, like “\\.” or
“\\+.” More on these special characters will be covered in
a future lesson.
Q: Separating Special Characters
Your next task involves the hiv_dat_clean_1 dataset.
Focus on the regimen column, which lists drug regimens
separated by a + sign. Your goal is to split this column
into three new columns: drug_1, drug_2, and
drug_3 using the separate() function. Pay
close attention to how you handle the + separator. Here’s
the column:
## # A tibble: 1,413 × 1
## regimen
## <chr>
## 1 AZT+3TC+NVP
## 2 TDF+3TC+EFV
## 3 TDF+3TC+EFV
## 4 TDF+3TC+EFV
## 5 TDF+3TC+EFV
## 6 AZT+3TC+NVP
## 7 TDF+3TC+EFV
## 8 AZT+3TC+NVP
## 9 AZT+3TC+NVP
## 10 TDF+3TC+EFV
## # ℹ 1,403 more rows7 Combining Strings with
paste()
The paste() function in R concatenates or joins together
character strings. This allows you to combine multiple strings into a
single string.
To combine two simple strings:
## [1] "Hello World"The default separator is a space, so this returns “Hello World”.
Let’s demonstrate how to use this on a dataset, with the IRS date data. First, we’ll separate the start date into individual columns:
irs_dates_separated <- # store for later use
irs_dates_1 %>%
separate(start_date_long, into = c("month", "day", "year"), sep = " ", remove = FALSE)
irs_dates_separated## # A tibble: 112 × 5
## village start_date_long month day year
## <chr> <chr> <chr> <chr> <chr>
## 1 Mess April 07 2014 April 07 2014
## 2 Nkombedzi April 22 2014 April 22 2014
## 3 B Compound May 13 2014 May 13 2014
## 4 D Compound May 13 2014 May 13 2014
## 5 Post Office May 13 2014 May 13 2014
## 6 Mangulenje May 15 2014 May 15 2014
## 7 Mangulenje Senior May 27 2014 May 27 2014
## 8 Old School May 27 2014 May 27 2014
## 9 Mwanza May 28 2014 May 28 2014
## 10 Alumenda June 18 2014 June 18 2014
## # ℹ 102 more rowsThen we can recombine day, month and year with
paste():
irs_dates_separated %>%
select(day, month, year) %>%
mutate(start_date_long_2 = paste(day, month, year))## # A tibble: 112 × 4
## day month year start_date_long_2
## <chr> <chr> <chr> <chr>
## 1 07 April 2014 07 April 2014
## 2 22 April 2014 22 April 2014
## 3 13 May 2014 13 May 2014
## 4 13 May 2014 13 May 2014
## 5 13 May 2014 13 May 2014
## 6 15 May 2014 15 May 2014
## 7 27 May 2014 27 May 2014
## 8 27 May 2014 27 May 2014
## 9 28 May 2014 28 May 2014
## 10 18 June 2014 18 June 2014
## # ℹ 102 more rowsThe sep argument specifies the separator between
elements. For a different separator, like a hyphen, we can write:
## # A tibble: 112 × 6
## village start_date_long month day year
## <chr> <chr> <chr> <chr> <chr>
## 1 Mess April 07 2014 April 07 2014
## 2 Nkombedzi April 22 2014 April 22 2014
## 3 B Compound May 13 2014 May 13 2014
## 4 D Compound May 13 2014 May 13 2014
## 5 Post Office May 13 2014 May 13 2014
## 6 Mangulenje May 15 2014 May 15 2014
## 7 Mangulenje Senior May 27 2014 May 27 2014
## 8 Old School May 27 2014 May 27 2014
## 9 Mwanza May 28 2014 May 28 2014
## 10 Alumenda June 18 2014 June 18 2014
## # ℹ 102 more rows
## # ℹ 1 more variable: start_date_long_2 <chr>To concatenate without spaces, we can set sep = "":
irs_dates_separated %>%
select(day, month, year) %>%
mutate(start_date_long_2 = paste(day, month, year, sep = ""))## # A tibble: 112 × 4
## day month year start_date_long_2
## <chr> <chr> <chr> <chr>
## 1 07 April 2014 07April2014
## 2 22 April 2014 22April2014
## 3 13 May 2014 13May2014
## 4 13 May 2014 13May2014
## 5 13 May 2014 13May2014
## 6 15 May 2014 15May2014
## 7 27 May 2014 27May2014
## 8 27 May 2014 27May2014
## 9 28 May 2014 28May2014
## 10 18 June 2014 18June2014
## # ℹ 102 more rowsOr we can use the paste0() function, which is equivalent
to paste(..., sep = ""):
irs_dates_separated %>%
select(day, month, year) %>%
mutate(start_date_long_2 = paste0(day, month, year))## # A tibble: 112 × 4
## day month year start_date_long_2
## <chr> <chr> <chr> <chr>
## 1 07 April 2014 07April2014
## 2 22 April 2014 22April2014
## 3 13 May 2014 13May2014
## 4 13 May 2014 13May2014
## 5 13 May 2014 13May2014
## 6 15 May 2014 15May2014
## 7 27 May 2014 27May2014
## 8 27 May 2014 27May2014
## 9 28 May 2014 28May2014
## 10 18 June 2014 18June2014
## # ℹ 102 more rowsLet’s try to combine paste() with some other string
functions to solve a realistic data problem. Consider the ID column in
the hiv_dat_messy_1 dataset:
## # A tibble: 1,413 × 1
## patient_id
## <chr>
## 1 pd-10037
## 2 pd-10537
## 3 pd-5489
## 4 id-5523
## 5 pd-4942
## 6 pd-4742
## 7 pd-10879
## 8 id-2885
## 9 pd-4861
## 10 pd-5180
## # ℹ 1,403 more rowsImagine we wanted to standardize these IDs to have the same number of characters. This is often a requirement for IDs (think about phone numbers, for instance).
To implement this, we can use separate() to split the
IDs into parts, then use paste() to recombine them into a
standardized format.
hiv_dat_messy_1 %>%
select(patient_id) %>% # for visibility
separate(patient_id, into = c("prefix", "patient_num"), sep = "-", remove = F) %>%
mutate(patient_num = str_pad(patient_num, width = 5, side = "left", pad = "0")) %>%
mutate(patient_id_padded = paste(prefix, patient_num, sep = "-"))## # A tibble: 1,413 × 4
## patient_id prefix patient_num patient_id_padded
## <chr> <chr> <chr> <chr>
## 1 pd-10037 pd 10037 pd-10037
## 2 pd-10537 pd 10537 pd-10537
## 3 pd-5489 pd 05489 pd-05489
## 4 id-5523 id 05523 id-05523
## 5 pd-4942 pd 04942 pd-04942
## 6 pd-4742 pd 04742 pd-04742
## 7 pd-10879 pd 10879 pd-10879
## 8 id-2885 id 02885 id-02885
## 9 pd-4861 pd 04861 pd-04861
## 10 pd-5180 pd 05180 pd-05180
## # ℹ 1,403 more rowsIn this example, patient_id is split into a prefix and a
number. The number is then padded with zeros to ensure consistent
length, and finally, the two parts are concatenated back together using
paste() with a hyphen as the separator. This process
standardizes the format of patient IDs.
Great work!
Q: Standardizing IDs in the
lima_messy_1 Dataset
In the lima_messy_1 dataset, the IDs are not
zero-padded, making them hard to sort.
For example, the ID pe-998 is at the top of the list
after sorting in descending order, which is not what we want.
lima_messy_1 %>%
select(id) %>%
arrange(desc(id)) # sort in descending order (highest IDs should be at the top)## # A tibble: 1,293 × 1
## id
## <chr>
## 1 pe-998
## 2 pe-996
## 3 pe-951
## 4 pe-900
## 5 pe-2347
## 6 pe-2337
## 7 pe-2335
## 8 pe-2333
## 9 pe-2331
## 10 pe-2329
## # ℹ 1,283 more rowsTry to fix this using a similar procedure to the one used for
hiv_dat_messy_1.
Your Task:
- Separate the ID into parts.
- Pad the numeric part for standardization.
- Recombine the parts using
paste(). - Resort the IDs in descending order. The highest ID should end in
2347
Q: Creating Summary Statements
Create a column containing summary statements combining
village, start_date_default, and
coverage_p from the irs dataset. The statement
should describe the spray coverage for each village.
Desired Output: “For village X, the spray coverage was Y% on Z date.”
Your Task: - Select the necessary columns from the
irs dataset. - Use paste() to create the
summary statement.
As we go through this lesson, remember that RStudio’s auto-complete can help you find functions in the stringr package.
Just type str_ and a list of stringr functions will pop
up. All stringr functions start with str_.
So instead of trying to memorize them all, you can use auto-complete as a reference when needed.
8 Subsetting strings with
str_sub
str_sub allows you to extract parts of a string based on
character positions. The basic syntax is
str_sub(string, start, end).
Example: Extracting the first 2 characters from patient IDs:
## [1] "ID" "ID"Or the first 5:
## [1] "ID123" "ID678"Negative values count backward from the end of the string. This is useful for extracting suffixes.
For example, to get the last 4 characters of patient IDs.
## [1] "-abc" "-def"Be sure to pause and understand what happened above.
When indices are outside the string length, str_sub
handles it gracefully without errors:
str_sub(patient_ids, 1, 30) # Safely returns the full string when the range exceeds the string length## [1] "ID12345-abc" "ID67890-def"In a data frame, we can use str_sub within
mutate(). For example, below we extract the year and month
from the start_date_default column and create a new column
called year_month:
irs %>%
select(start_date_default) %>%
mutate(year_month = str_sub(start_date_default, start = 1, end = 7))## # A tibble: 112 × 2
## start_date_default year_month
## <date> <chr>
## 1 2014-04-07 2014-04
## 2 2014-04-22 2014-04
## 3 2014-05-13 2014-05
## 4 2014-05-13 2014-05
## 5 2014-05-13 2014-05
## 6 2014-05-15 2014-05
## 7 2014-05-27 2014-05
## 8 2014-05-27 2014-05
## 9 2014-05-28 2014-05
## 10 2014-06-18 2014-06
## # ℹ 102 more rowsWrap up
Congratulations on reaching the end of this lesson! You’ve learned about strings in R and various functions to manipulate them effectively.
The table below gives a quick recap of the key functions we covered. Remember, you don’t need to memorize all these functions. Knowing they exist and how to look them up (like using Google) is more than enough for practical applications.
| Function | Description | Example | Example Output |
str_to_upper() |
Convert characters to uppercase | str_to_upper("hiv") |
“HIV” |
str_to_lower() |
Convert characters to lowercase | str_to_lower("HIV") |
“hiv” |
str_to_title() |
Convert first character of each word to uppercase | str_to_title("hiv awareness") |
“Hiv Awareness” |
str_trim() |
Remove whitespace from start & end | str_trim(" hiv ") |
“hiv” |
str_squish() |
Remove whitespace from start & end and reduce internal spaces | str_squish(" hiv cases ") |
“hiv cases” |
str_pad() |
Pad a string to a fixed width | str_pad("45", width = 5) |
“00045” |
str_wrap() |
Wrap a string to a given width (for formatting output) | str_wrap("HIV awareness", width = 5) |
“HIV” |
str_split() |
Split elements of a character vector | str_split("Hello-World", "-") |
c(“Hello”, “World”) |
paste() |
Concatenate vectors after converting to character | paste("Hello", "World") |
“Hello World” |
str_sub() |
Extract and replace substrings from a character vector | str_sub("HelloWorld", 1, 4) |
“Hell” |
separate() |
Separate a character column into multiple columns |
|
|b |c ||Hello |World | |
Note that while these functions cover common tasks such as string standardization, splitting and joining strings, this introduction only scratches the surface of what’s possible with the {stringr} package. If you work with a lot of raw text data, you may want to do further exploring on the stringr website.
Answer Key
Q: Error Spotting in String Definitions
ex_a: Correct.ex_b: Correct.ex_c: Error. Corrected version:ex_c <- "They've been \"best friends\" for years."ex_d: Error. Corrected version:ex_d <- 'Jane\'s diary'ex_e: Error. Close quote missing. Corrected version:ex_e <- "It's a sunny day!"
Q: Cleaning Patient Name Data
Q: Standardizing Drug Codes
Q: Wrapping Medical Instructions
instructions <- "Take two tablets daily after meals. If symptoms persist for more than three days, consult your doctor immediately. Do not take more than the recommended dose. Keep out of reach of children."
# Wrap instructions
wrapped_instructions <- str_wrap(instructions, width = 50)
ggplot(data.frame(x = 1, y = 1), aes(x, y, label = wrapped_instructions)) +
geom_label() +
theme_void()Q: Formatting a Tuberculosis Dataset
The steps to clean the lima_messy dataset would
involve:
lima_clean <- lima_messy %>%
mutate(
marital_status = str_squish(str_to_title(marital_status)), # Clean and standardize marital_status
sex = str_squish(str_to_upper(sex)), # Clean and standardize sex
age = str_squish(str_to_lower(age)) # Clean and standardize age
)
lima_clean %>%
select(marital_status, sex, age) %>%
tbl_summary()Then, use the tbl_summary() function to create the
summary table.
Q: Wrapping Axis Labels in a Plot
Q: Splitting Age Range Strings
Q: Creating Summary Statements
Contributors
The following team members contributed to this lesson:
KENE DAVID NWOSU
Passionate about world improvement