What is the way to wrap ggplot call inside a function and pass function arguments to ggplot's a-CodePudding

Consider the minimal working example

library(tidyverse)

myplot <- function(df, x, y) {
  df |>
    ggplot(aes(x, y))  
    geom_point()
}

iris |>
  myplot(Petal.Width, Petal.Length)

It raises Error in FUN(X[[i]], ...) : object 'Petal.Width' not found. How do I delay the evaluation of Petal.Width and Petal.Length to ggplot?

CodePudding user response：

One option would be to use curly-curly {{ which replaces the old pattern of using bang-bang !! enquo. For more info see What is data-masking and why do I need {{? and the ggplot2 book for using {{ in a ggplot2 context:

library(tidyverse)

myplot <- function(df, x, y) {
  df |>
    ggplot(aes({{ x }}, {{ y }}))  
    geom_point()
}

iris |>
  myplot(Petal.Width, Petal.Length)

CodePudding user response：

1) dot dot dot R's dot dot dot works here:

myplot <- function(df, ...) {
  df |>
    ggplot(aes(...))  
    geom_point()
}

iris |>
  myplot(Petal.Width, Petal.Length)

2) aes_string Another possibility is to use standard evaluation with aes_string.

myplot <- function(df, x, y) {
  df |>
    ggplot(aes_string(x, y))  
    geom_point()
}

iris |>
  myplot("Petal.Width", "Petal.Length")

3) formula A variation of (2) which also uses aes_string is to pass a formula. Use all.vars to extract the variable names.

myplot <- function(df, formula) {
  v <- all.vars(formula)
  df |>
    ggplot(aes_string(v[2], v[1]))  
    geom_point()
}

iris |>
  myplot(Petal.Length ~ Petal.Width)

4) plot/xyplot plot (base R) and lattice::xyplot arguments are the same as (3) except for order giving trivial implementations.

library(lattice)

myplot <- function(df, formula) {
  plot(formula, df)
}

iris |>
  myplot(Petal.Length ~ Petal.Width)


library(lattice)

myplot <- function(df, formula) {
  xyplot(formula, df)
}

iris |>
  myplot(Petal.Length ~ Petal.Width)

CodePudding user response：

Using non-standard evaluation https://www.r-bloggers.com/2019/07/bang-bang-how-to-program-with-dplyr/

library(tidyverse)

myplot <- function(df, x, y) {
  df |>
    ggplot(aes(!!x, !!y))  
    geom_point()
}



iris |> 
  myplot(quo(Petal.Width), quo(Petal.Length))