ggplot, facets, and scale labelling-CodePudding

I am producing a facet chart like this

suppressPackageStartupMessages({
  library(tidyverse)
  library(scales)
})

set.seed(123456)
test_df <-
  tibble(t=1:10,
         x=100 runif(10,-10,10),
         y=10 runif(10,-4,4),
         p=y/x)

test_df |>
  pivot_longer(cols=x:p,
               names_to = "var",
               values_to = "value") |>
  mutate(var=factor(var,levels=c("x","y","p"))) ->
test_df_long

test_df_long |>
  ggplot(aes(x=t,y=value))  
  geom_line()  
  scale_y_continuous(limits=c(0,NA))  
  facet_wrap(vars(var),ncol=1,
             scales = "free_y")

I would like to change the formatting of the labels for the p facet to percent.

I have not found a way to use different labelling functions to different facets.

Can it be done?

CodePudding user response：

You can use the package facetscales like this:

devtools::install_github("zeehio/facetscales")
library(facetscales)

scales_y <- list(
  x = scale_y_continuous(),
  y = scale_y_continuous(),
  p = scale_y_continuous(labels = percent_format())
)

test_df_long |>
  ggplot(aes(x=t,y=value))  
  geom_line()  
  scale_y_continuous(limits=c(0,NA))  
  facet_grid_sc(rows = vars(var), scales = list(y = scales_y))

Output:

CodePudding user response：

You could use ggh4x::facetted_pos_scales() to control position scales of individual facets. If you just need to adjust a single scale, you could use the formula notation as below. (Disclaimer: I'm the author of ggh4x)

suppressPackageStartupMessages({
  library(tidyverse)
  library(scales)
})

set.seed(123456)
test_df <-
  tibble(t=1:10,
         x=100 runif(10,-10,10),
         y=10 runif(10,-4,4),
         p=y/x)

test_df |>
  pivot_longer(cols=x:p,
               names_to = "var",
               values_to = "value") |>
  mutate(var=factor(var,levels=c("x","y","p"))) ->
  test_df_long

test_df_long |>
  ggplot(aes(x=t,y=value))  
  geom_line()  
  scale_y_continuous(limits=c(0,NA))  
  facet_wrap(vars(var),ncol=1,
             scales = "free_y")  
  ggh4x::facetted_pos_scales(
    y = var == "p" ~ scale_y_continuous(labels = percent_format())
  )

^{Created on 2022-05-11 by the reprex package (v2.0.1)}