fct_relevel

In this document, I will introduce the fct_relevel function and show what it’s used for.

fct_relevel is part of the forcats R package.

I will be using icu_Hoda_edited dataset adapted from 513 BSTA class.

I am interested in “race” in this dataframe.

tabyl (icu_Hoda_edited$race)
##  icu_Hoda_edited$race   n percent
##                 Black  15   0.075
##                 Other  10   0.050
##                 White 175   0.875
class(icu_Hoda_edited$race)
## [1] "character"

I factorize it then check for its levels.

icu_Hoda_edited <-icu_Hoda_edited%>% mutate(race= factor(race))
class(icu_Hoda_edited$race)
## [1] "factor"
levels(icu_Hoda_edited$race)
## [1] "Black" "Other" "White"

The default order for levels with factor() is ALPHABETICAL.

What is fct_relevel used for?

The function simply is used to reorder factor levels by hand when we want to specify the reference level.

(e.g. when plotting, or when performing a regression analysis)

If we want to make comparisons relative to “White” race, we need to move it to the front by using fct_relevel().

icu_Hoda_edited <- icu_Hoda_edited%>% 
  mutate(race = race %>%
           fct_relevel("White"))

levels(icu_Hoda_edited$race)
## [1] "White" "Black" "Other"

“White” has moved to to be the first level (reference).

The rest of the other levels remain ordered alphabetically.

If we desire to move “Other” to be the second level, we add it to the argument.

icu_Hoda_edited <- icu_Hoda_edited%>% 
  mutate(race = race %>%
           fct_relevel("White","Other"))

levels(icu_Hoda_edited$race)
## [1] "White" "Other" "Black"

Similar functions:

levels()

race_factor <- factor(race_character,levels = c(“White”, “Other”, “Black”))

  • Useful when creating a factor.

  • May not be helpful if we have a lot of levels because we will need to write them all down - long code!

relevel()

race_factor<-relevel(race_factor, ref=“White”)

  • Useful for specifying only the reference (1st level)

Is it helpful?

YES! fct_relevel is useful if we want to reorder the levels in a factor to specify the reference level.

Helps us “pull” a level (or more) to the front relative to all other levels. The rest of the levels remain unchanged and follow according to their alphabetical order.

Source: R for Health Data Science Ewen Harrison and Riinu Pius 2021-01-15 https://argoshare.is.ed.ac.uk/healthyr_book/chap08-h2-fct-relevel.html