What is the Tidyverse?
The tidyverse is an opinionated collection of R packages designed for data science. All packages share an underlying design philosophy, grammar, and data structures.
www.tidyverse.org
- There is nothing that can be done using the tidyverse which cannot be accomplished using base R or other packages.
- I prefer these packages because I found a significant leap forward in the robustness of my code, quicker debugging time, and quicker to perform checks on my data.
Further, I recommend the book R for Data science which is freely available online at https://r4ds.had.co.nz/
What is the Tidyverse?
The packages within the tidyverse are categorized into two parts: 8 core packages and 15 non-core packages.
There is a higher level management package called tidyverse which helps maintain the whole collection of packages. All 23 packages can be installed with a single function.
install.packages("tidyverse")
The Tidyverse
The core packages are likely to be used each time you sit down to write code, and loading JUST the core 8 packages can be done with a single function.
library(tidyverse)## -- Attaching packages ------------------------------ tidyverse 1.3.0 --## v ggplot2 3.3.2 v purrr 0.3.4## v tibble 3.0.1 v dplyr 1.0.2## v tidyr 1.1.0 v stringr 1.4.0## v readr 1.3.1 v forcats 0.5.0## -- Conflicts --------------------------------- tidyverse_conflicts() --## x dplyr::filter() masks stats::filter()## x dplyr::group_rows() masks kableExtra::group_rows()## x dplyr::lag() masks stats::lag()
The Tidyverse
The core packages are likely to be used each time you sit down to write code, and loading JUST the core 8 packages can be done with a single function.
library(tidyverse)## -- Attaching packages ------------------------------ tidyverse 1.3.0 --## v ggplot2 3.3.2 v purrr 0.3.4## v tibble 3.0.1 v dplyr 1.0.2## v tidyr 1.1.0 v stringr 1.4.0## v readr 1.3.1 v forcats 0.5.0## -- Conflicts --------------------------------- tidyverse_conflicts() --## x dplyr::filter() masks stats::filter()## x dplyr::group_rows() masks kableExtra::group_rows()## x dplyr::lag() masks stats::lag()
Dplyr
Dplyr provides a simple set of “verb” functions that make basic data manipulation easier:
filter()- creates a subset of the data by extracting rows which meet certain criteria
select()- creates a subset of the data by extracting a set of columns we identify by name
mutate()- creates/transforms variables using pre-existing variables and adds them to the end of our tibble
Dplyr
Dplyr provides a simple set of “verb” functions that make basic data manipulation easier:
filter()- creates a subset of the data by extracting rows which meet certain criteria
select()- creates a subset of the data by extracting a set of columns we identify by name
mutate()- creates/transforms variables using pre-existing variables and adds them to the end of our tibble
group_by()- adds grouping meta-data to the data
Dplyr
Dplyr provides a simple set of “verb” functions that make basic data manipulation easier:
filter()- creates a subset of the data by extracting rows which meet certain criteria
select()- creates a subset of the data by extracting a set of columns we identify by name
mutate()- creates/transforms variables using pre-existing variables and adds them to the end of our tibble
group_by()- adds grouping meta-data to the data
summarise()- applies summary functions to columns in our data and returns a tibble of the results
Dplyr
This link will download the cheat sheet:
https://www.rstudio.org/links/data_transformation_cheat_sheet
This link has all of the help documentation and listing of available funcions:
https://www.rdocumentation.org/packages/dplyr/versions/0.7.8
Magrittr
The pipe operator %>% is a very useful way of writing human-readable code, while avoiding intermediate objects within R to hold results.
This operator can be read as the phrase "and then...". It takes the results from the left side of the operator and inserts it as the FIRST argument of the right side of the operator.
# original function callf(x,y)# becomesx %>% f(y)# original function callfilter(mtcars, cyl == 4)# becomesmtcars %>% filter(cyl == 4)Magrittr
The pipe operator %>% is a very useful way of writing human-readable code, while avoiding intermediate objects within R to hold results.
This operator can be read as the phrase "and then...". It takes the results from the left side of the operator and inserts it as the FIRST argument of the right side of the operator.
# original function callf(x,y)# becomesx %>% f(y)
Data Cleaning/Cleansing
Data Quality can be broken down into a few categories:
- Validity
- In the sense of either valid or invalid, and not the degree to which a measurement corresponds to a true (possibly unknown) value.
- What is your first name? 3 would not be a valid response
- In the sense of either valid or invalid, and not the degree to which a measurement corresponds to a true (possibly unknown) value.
- Accuracy
- This is what psychologists typically refer to as validity or closeness of a value to a known standard or true value.
Data Cleaning/Cleansing
Data Quality can be broken down into a few categories:
- Validity
- In the sense of either valid or invalid, and not the degree to which a measurement corresponds to a true (possibly unknown) value.
- What is your first name? 3 would not be a valid response
- In the sense of either valid or invalid, and not the degree to which a measurement corresponds to a true (possibly unknown) value.
- Accuracy
- This is what psychologists typically refer to as validity or closeness of a value to a known standard or true value.
- Completeness
- The degree to which all required measures are known.
- For us, this pertains to missing data
Data Cleaning/Cleansing
Data Quality can be broken down into a few categories:
- Validity
- In the sense of either valid or invalid, and not the degree to which a measurement corresponds to a true (possibly unknown) value.
- What is your first name? 3 would not be a valid response
- In the sense of either valid or invalid, and not the degree to which a measurement corresponds to a true (possibly unknown) value.
- Accuracy
- This is what psychologists typically refer to as validity or closeness of a value to a known standard or true value.
- Completeness
- The degree to which all required measures are known.
- For us, this pertains to missing data
- Consistency
- Pertains to incompatabilities within the data itself or as can be corroborated with other existing data
Data Cleaning/Cleansing
Data Quality can be broken down into a few categories:
- Validity
- In the sense of either valid or invalid, and not the degree to which a measurement corresponds to a true (possibly unknown) value.
- What is your first name? 3 would not be a valid response
- In the sense of either valid or invalid, and not the degree to which a measurement corresponds to a true (possibly unknown) value.
- Accuracy
- This is what psychologists typically refer to as validity or closeness of a value to a known standard or true value.
- Completeness
- The degree to which all required measures are known.
- For us, this pertains to missing data
- Consistency
- Pertains to incompatabilities within the data itself or as can be corroborated with other existing data
- Uniformity
- Are all the values within a set of measures specified using the same units of measurement (inches vs cm)
Importing Data
data_superhero <- read_csv(file = here("data", "heroes_information.csv"))## Warning: Missing column names filled in: 'X1' [1]## Parsed with column specification:## cols(## X1 = col_double(),## name = col_character(),## Gender = col_character(),## `Eye color` = col_character(),## Race = col_character(),## `Hair color` = col_character(),## Height = col_double(),## Publisher = col_character(),## `Skin color` = col_character(),## Alignment = col_character(),## Weight = col_double()## )
Importing Data
data_superhero <- read_csv(file = here("data", "heroes_information.csv"))## Warning: Missing column names filled in: 'X1' [1]## Parsed with column specification:## cols(## X1 = col_double(),## name = col_character(),## Gender = col_character(),## `Eye color` = col_character(),## Race = col_character(),## `Hair color` = col_character(),## Height = col_double(),## Publisher = col_character(),## `Skin color` = col_character(),## Alignment = col_character(),## Weight = col_double()## )spec(data_superhero)
Importing Data
data_superhero <- read_csv(file = here("data", "heroes_information.csv"))## Warning: Missing column names filled in: 'X1' [1]## Parsed with column specification:## cols(## X1 = col_double(),## name = col_character(),## Gender = col_character(),## `Eye color` = col_character(),## Race = col_character(),## `Hair color` = col_character(),## Height = col_double(),## Publisher = col_character(),## `Skin color` = col_character(),## Alignment = col_character(),## Weight = col_double()## )spec(data_superhero)
X1?
Let's take a peak at the mysterious variable X1
glimpse(data_superhero)## Rows: 734## Columns: 11## $ X1 <dbl> 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, ...## $ name <chr> "A-Bomb", "Abe Sapien", "Abin Sur", "Abomination", "Ab...## $ Gender <chr> "Male", "Male", "Male", "Male", "Male", "Male", "Male"...## $ `Eye color` <chr> "yellow", "blue", "blue", "green", "blue", "blue", "bl...## $ Race <chr> "Human", "Icthyo Sapien", "Ungaran", "Human / Radiatio...## $ `Hair color` <chr> "No Hair", "No Hair", "No Hair", "No Hair", "Black", "...## $ Height <dbl> 203, 191, 185, 203, -99, 193, -99, 185, 173, 178, 191,...## $ Publisher <chr> "Marvel Comics", "Dark Horse Comics", "DC Comics", "Ma...## $ `Skin color` <chr> "-", "blue", "red", "-", "-", "-", "-", "-", "-", "-",...## $ Alignment <chr> "good", "good", "good", "bad", "bad", "bad", "good", "...## $ Weight <dbl> 441, 65, 90, 441, -99, 122, -99, 88, 61, 81, 104, 108,...
X1?
Let's take a peak at the mysterious variable X1
glimpse(data_superhero)## Rows: 734## Columns: 11## $ X1 <dbl> 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, ...## $ name <chr> "A-Bomb", "Abe Sapien", "Abin Sur", "Abomination", "Ab...## $ Gender <chr> "Male", "Male", "Male", "Male", "Male", "Male", "Male"...## $ `Eye color` <chr> "yellow", "blue", "blue", "green", "blue", "blue", "bl...## $ Race <chr> "Human", "Icthyo Sapien", "Ungaran", "Human / Radiatio...## $ `Hair color` <chr> "No Hair", "No Hair", "No Hair", "No Hair", "Black", "...## $ Height <dbl> 203, 191, 185, 203, -99, 193, -99, 185, 173, 178, 191,...## $ Publisher <chr> "Marvel Comics", "Dark Horse Comics", "DC Comics", "Ma...## $ `Skin color` <chr> "-", "blue", "red", "-", "-", "-", "-", "-", "-", "-",...## $ Alignment <chr> "good", "good", "good", "bad", "bad", "bad", "good", "...## $ Weight <dbl> 441, 65, 90, 441, -99, 122, -99, 88, 61, 81, 104, 108,...n_distinct(data_superhero$X1)## [1] 734
Pro-tip: Unique Identifiers
If the data set does NOT have a unique identifier for each row (or these identifiers are stored as row names), then I highly recommend you create one.
data_superhero %>% rownames_to_column(var = "id")# ordata_superhero %>% add_column(id = 1:nrow(data_superhero), .before = 1)I prefer the first because row names in R are already constrained to be unique.
Importing Data: Consistent Variable Names
# Import data using a relative file pathdata_superhero <- read_csv(file = here("data", "heroes_information.csv")) %>% # rename the first column as "id" rename("id" = X1) %>% # rename variables to make them easier to use rename_with(.fn = snakecase::to_snake_case)# check that names were fixed properlycolnames(data_superhero)## [1] "id" "name" "gender" "eye_color" "race" ## [6] "hair_color" "height" "publisher" "skin_color" "alignment" ## [11] "weight"
Importing Data: Data-Type Constraints
Data-type constraints is next thing I check once I load data into R.
The readr package uses the first 1000 rows in a dataset to guess the most appropriate atomic type for each column in your dataset (they can be changed later):
- character
- "hello world"
- double
- 3.1459
- integer
- 4L
- logical
- TRUE, FALSE
Example of Boundary Constraints
Dplyr has a between() function which can be helpful when used in conjunction with the negation operator !
# What if we were to assume that any height larger than 300cm is abnormally large?data_superhero %>% filter(!between(x = height, left = 0, right = 300))## # A tibble: 225 x 11## id name gender eye_color race hair_color height publisher skin_color## <dbl> <chr> <chr> <chr> <chr> <chr> <dbl> <chr> <chr> ## 1 4 Abra~ Male blue Cosm~ Black -99 Marvel C~ - ## 2 6 Adam~ Male blue - Blond -99 NBC - He~ - ## 3 14 Alex~ Male - Human - -99 Wildstorm - ## 4 15 Alex~ Male - - - -99 NBC - He~ - ## 5 18 Alla~ Male - - - -99 Wildstorm - ## 6 21 Ando~ Male - - - -99 NBC - He~ - ## 7 23 Angel Male - Vamp~ - -99 Dark Hor~ - ## 8 26 Ange~ Female - - - -99 Image Co~ - ## 9 32 Anti~ Male - - - -99 Image Co~ - ## 10 35 Aqua~ Male blue - Blond -99 DC Comics - ## # ... with 215 more rows, and 2 more variables: alignment <chr>, weight <dbl>
Example of Boundary Constraints
Dplyr has a between() function which can be helpful when used in conjunction with the negation operator !
# What if we were to assume that any height larger than 300cm is abnormally large?data_superhero %>% filter(!between(x = height, left = 0, right = 300))## # A tibble: 225 x 11## id name gender eye_color race hair_color height publisher skin_color## <dbl> <chr> <chr> <chr> <chr> <chr> <dbl> <chr> <chr> ## 1 4 Abra~ Male blue Cosm~ Black -99 Marvel C~ - ## 2 6 Adam~ Male blue - Blond -99 NBC - He~ - ## 3 14 Alex~ Male - Human - -99 Wildstorm - ## 4 15 Alex~ Male - - - -99 NBC - He~ - ## 5 18 Alla~ Male - - - -99 Wildstorm - ## 6 21 Ando~ Male - - - -99 NBC - He~ - ## 7 23 Angel Male - Vamp~ - -99 Dark Hor~ - ## 8 26 Ange~ Female - - - -99 Image Co~ - ## 9 32 Anti~ Male - - - -99 Image Co~ - ## 10 35 Aqua~ Male blue - Blond -99 DC Comics - ## # ... with 215 more rows, and 2 more variables: alignment <chr>, weight <dbl>We could insert code similar to this into our chain, but I recommend not subsetting out whole rows of code at this point. This is akin to list-wise deletion. Many modeling functions in R will perform listwise deletion automatically (unless you tell it otherwise). Depending on how you want to handle missing our outlying or influential cases, you might need to retain as much of your data as possible.
Range Constraints
p <- data_superhero %>% ggplot(aes(x = height)) + geom_histogram(bins = 20) + geom_vline(aes(xintercept = 0), color = "red")plotly::ggplotly(p)- Start with the data, then ...
- create a canvas with height on the x-axis, then ...
- generate a histogram, then ..
- add a vertical line at my boundary constraint
Range Constraints
p <- data_superhero %>% ggplot(aes(x = weight, y = height)) + geom_point() + geom_hline(yintercept = 0, color = "red") + geom_vline(xintercept = 0, color = "red")plotly::ggplotly(p)- Start with the data, then ...
- create a canvas with height on the y-axis and weight on the x-axis, then ...
- generate a scatterplot, then ..
- add a vertical and horizontal lines at my boundary constraints
We found a problem, now what?
We know that we have cases with negative height and weight, so what should we do about?
# examine negative or zero height valuesdata_superhero %>% filter(height <= 0) %>% count(height)## # A tibble: 1 x 2## height n## <dbl> <int>## 1 -99 217# examine negative or zero weight valuesdata_superhero %>% filter(weight <= 0) %>% count(weight)## # A tibble: 1 x 2## weight n## <dbl> <int>## 1 -99 237
We found a problem, now what?
That suspicious value of -99 was probably a way of encoding missing data, so we should update our cleaning script with two things.
- We insert our
mutate()we just created into our data pipeline (even if the negatives will be gone soon)- Remember to invert the operators to exclude the negative values
- We should convert these values into
NAso that R treats them appropriately.
data_superhero %>% mutate(height = na_if(height, y= -99), weight = na_if(weight, y = -99)) %>% View()
Was the problem fixed?
## # A tibble: 734 x 11## id name gender eye_color race hair_color height publisher skin_color## <dbl> <chr> <chr> <chr> <chr> <chr> <dbl> <chr> <chr> ## 1 0 A-Bo~ Male yellow Human No Hair 203 Marvel C~ - ## 2 1 Abe ~ Male blue Icth~ No Hair 191 Dark Hor~ blue ## 3 2 Abin~ Male blue Unga~ No Hair 185 DC Comics red ## 4 3 Abom~ Male green Huma~ No Hair 203 Marvel C~ - ## 5 4 Abra~ Male blue Cosm~ Black NA Marvel C~ - ## 6 5 Abso~ Male blue Human No Hair 193 Marvel C~ - ## 7 6 Adam~ Male blue - Blond NA NBC - He~ - ## 8 7 Adam~ Male blue Human Blond 185 DC Comics - ## 9 8 Agen~ Female blue - Blond 173 Marvel C~ - ## 10 9 Agen~ Male brown Human Brown 178 Marvel C~ - ## # ... with 724 more rows, and 2 more variables: alignment <chr>, weight <dbl>
Pro-Tip: Multiple Mutates
Mutate if a column meets a particular criterion
data_superhero %>% # replace -99 with NA for all double variables mutate_if(.predicate = is_double, .funs = na_if, y = -99)Mutate at all specific columns
data_superhero %>% # replace -99 with NA for all double variables mutate_at(.vars = vars(weight,height), .funs = na_if, y = -99)
Our Data Pipeline So Far
# Start by importing the data# I'm using the here::here() function to make the file path relative to the project directorydata_superhero <- read_csv(file = here("data", "heroes_information.csv")) %>% # rename the first column as "id" rename("id" = X1) %>% # rename variables to make them easier to use rename_with(.fn = to_snake_case) %>% # replace -99 with NA mutate(height = na_if(height, y= -99), weight = na_if(weight, y = -99))- We imported our data (using a relative path), then ...
- renamed the
X1variable toid, then ... - fixed variable names, then ...
- replaced values of -99 with
NAfor height and weight
Prepping Categorical Data
For some categorical data we want to remove accidental groups in data that could be the result of typographical or data-entry errors.
Consider the following chunk of code:
# create example dataundergrad_year <- c("First","Second","Third","second")# convert to a factor without explicitly setting the levelsundergrad_year_fct <- parse_factor(undergrad_year)undergrad_year_fct## [1] First Second Third second## Levels: First Second Third secondWe want to be sure that when we clean up certain variables, we are not losing potentially important aspects of our data.
So, let's take a peak at our data to determine which variables should be cleaned to prevent case-sensitivity related errors.
Cleaning Character Data
Before we clean up character data, we need to carefully consider whether the action we take cleaning could result in a loss of information.
data_superhero %>% mutate_at(.vars = c("gender", "eye_color", "race", "hair_color", "skin_color", "alignment", "publisher"), .funs = str_to_lower) %>% glimpse()## Rows: 734## Columns: 11## $ id <dbl> 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16...## $ name <chr> "A-Bomb", "Abe Sapien", "Abin Sur", "Abomination", "Abra...## $ gender <chr> "male", "male", "male", "male", "male", "male", "male", ...## $ eye_color <chr> "yellow", "blue", "blue", "green", "blue", "blue", "blue...## $ race <chr> "human", "icthyo sapien", "ungaran", "human / radiation"...## $ hair_color <chr> "no hair", "no hair", "no hair", "no hair", "black", "no...## $ height <dbl> 203, 191, 185, 203, -99, 193, -99, 185, 173, 178, 191, 1...## $ publisher <chr> "marvel comics", "dark horse comics", "dc comics", "marv...## $ skin_color <chr> "-", "blue", "red", "-", "-", "-", "-", "-", "-", "-", "...## $ alignment <chr> "good", "good", "good", "bad", "bad", "bad", "good", "go...## $ weight <dbl> 441, 65, 90, 441, -99, 122, -99, 88, 61, 81, 104, 108, 9...
Cleaning Character Data
Before we clean up character data, we need to carefully consider whether the action we take cleaning could result in a loss of information.
data_superhero %>% mutate_at(.vars = c("gender", "eye_color", "race", "hair_color", "skin_color", "alignment", "publisher"), .funs = str_to_lower) %>% glimpse()## Rows: 734## Columns: 11## $ id <dbl> 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16...## $ name <chr> "A-Bomb", "Abe Sapien", "Abin Sur", "Abomination", "Abra...## $ gender <chr> "male", "male", "male", "male", "male", "male", "male", ...## $ eye_color <chr> "yellow", "blue", "blue", "green", "blue", "blue", "blue...## $ race <chr> "human", "icthyo sapien", "ungaran", "human / radiation"...## $ hair_color <chr> "no hair", "no hair", "no hair", "no hair", "black", "no...## $ height <dbl> 203, 191, 185, 203, -99, 193, -99, 185, 173, 178, 191, 1...## $ publisher <chr> "marvel comics", "dark horse comics", "dc comics", "marv...## $ skin_color <chr> "-", "blue", "red", "-", "-", "-", "-", "-", "-", "-", "...## $ alignment <chr> "good", "good", "good", "bad", "bad", "bad", "good", "go...## $ weight <dbl> 441, 65, 90, 441, -99, 122, -99, 88, 61, 81, 104, 108, 9...
Profiling Categorical Data
So far we have cleaned/checked id and opted to leave name, let's continue by taking a peak at gender and alignment.
fct_count(data_superhero$gender)## # A tibble: 3 x 2## f n## <fct> <int>## 1 - 29## 2 female 200## 3 male 505fct_count(data_superhero$alignment)## # A tibble: 4 x 2## f n## <fct> <int>## 1 - 7## 2 bad 207## 3 good 496## 4 neutral 24There is a bizarre category with a value of "-", this probably indicates missing data for the character data within our dataset.
So let's go back to our cleaning script and slightly alter the read_csv() function.
Missing Character Data
The read_csv() function has many additional arguments which can alter how the data is imported into R.
We can automatically insert NA values when it encounters specific character values via the "na" argument.
data_superhero <- read_csv(file = here("data", "heroes_information.csv"), na = c("NA","","-","."))
Explicit Coercion
If you want more control over the coercion of your data into other atomic types or objects, then I recommend building a chain of functions. Testing each as you go along.
I recommend the readr::parse_*() functions for coercion.
data_superhero <- data_superhero %>% mutate(gender_fct = parse_factor(gender, levels = c("female","male")), alignment_fct = parse_factor(alignment, levels = c("good","neutral","bad")))# example of values with no level for them to belong toparse_factor(c("male","m","female"), levels = c("male","female"))## Warning: 1 parsing failure.## row col expected actual## 2 -- value in level set mReminder: problems()
Keep in mind that problems() comes from the readr package, so it only tracks problems from importing functions using the form readr::read_*() or coercion functions using the form readr::parse_*() from the readr package.
# check the whole dataset for parsing problems arising form importproblems(data_superhero)# check a single variable after using a parse_*() function for exlplicit coercionproblems(data_superhero$gender_fct)
eye_color
Let's keep going to see what else we can discover through profiling our data .
fct_count(data_superhero$eye_color) %>% arrange(f) %>% View()## # A tibble: 23 x 2## f n## <fct> <int>## 1 amber 2## 2 black 23## 3 blue 225## 4 blue / white 1## 5 bown 1## 6 brown 126## 7 gold 3## 8 green 73## 9 green / blue 1## 10 grey 6## # ... with 13 more rows
Edit eye_color
# replace all instances of "bown" eye_color with "brown"data_superhero %>% mutate(eye_color = if_else(condition = eye_color == "bown", true = "brown", false = eye_color)) %>% # Check that there are no cases of "bown" remaining filter(eye_color == "bown")## # A tibble: 0 x 13## # ... with 13 variables: id <dbl>, name <chr>, gender <chr>, eye_color <chr>,## # race <chr>, hair_color <chr>, height <dbl>, publisher <chr>,## # skin_color <chr>, alignment <chr>, weight <dbl>, gender_fct <fct>,## # alignment_fct <fct>- Start with our Cleaned (so far) data, then ...
mutate()theeye_colorvariable, then ...- check each value of eye_color check if our condition is true
- If true, then replace that value with "brown"
- If false, keep the original value of
eye_color
- check each value of eye_color check if our condition is true
- keep only the rows for superheroes with an eye_color exactly equal to "bown"
Visualize skin_color
p <- fct_count(data_superhero$skin_color) %>% arrange(f) %>% tidyr::drop_na(f) %>% ggplot(aes(x = f, y = n)) + geom_col() + coord_flip() + theme(text = element_text(size = 18))# generate interactive plotplotly::ggplotly(p)Let's try again, but let's insert another function from the tidyr package to drop rows which have a missing level.
Edit skin_color
# replace all instances of "gray" skin_color with "grey"data_superhero %>% mutate(skin_color = if_else(condition = skin_color == "gray", true = "grey", false = skin_color)) %>%# Check that there are no cases of "gray" remaining filter(skin_color == "gray")## # A tibble: 0 x 13## # ... with 13 variables: id <dbl>, name <chr>, gender <chr>, eye_color <chr>,## # race <chr>, hair_color <chr>, height <dbl>, publisher <chr>,## # skin_color <chr>, alignment <chr>, weight <dbl>, gender_fct <fct>,## # alignment_fct <fct>Edit skin_color
# replace all instances of "gray" skin_color with "grey"data_superhero %>% mutate(skin_color = if_else(condition = skin_color == "gray", true = "grey", false = skin_color)) %>%# Check that there are no cases of "gray" remaining filter(skin_color == "gray")## # A tibble: 0 x 13## # ... with 13 variables: id <dbl>, name <chr>, gender <chr>, eye_color <chr>,## # race <chr>, hair_color <chr>, height <dbl>, publisher <chr>,## # skin_color <chr>, alignment <chr>, weight <dbl>, gender_fct <fct>,## # alignment_fct <fct>