facet_grid() supporting a different scale per each facet.

[zeehio]

Motivation

The ability to use different scales in different facets has been a long time feature request (e.g. two years ago in #1613) and back in 2010 here.

The main reason to ask for this feature is to be able to compare different magnitudes, as would be desired for instance in this plot:

This kind of plots are usual in scientific environments (for instance for comparing concentrations, #1613 (comment)) and currently require a lot of fine tuning, using grid.arrange and playing with legends to make sure each subplot is aligned.

Example of use:

  library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
  library(scales)
  library(ggplot2)
# Simple tidy data frame, with 4 subjects. We measure three variables on each
# subject (namely: SomeValue, Percent and Scientific).
mydf <- dplyr::data_frame(Subject=rep(c("Alice", "Bob", "Charlie", "Diane"), each=3),
                          Magnitude=rep(c("SomeValue", "Percent", "Scientific"), times=4),
                          Value=c(c(170,0.6,2.7E-4),
                                  c(180, 0.8, 2.5E-4),
                                  c(160, 0.71, 3.2E-4),
                                  c(159, 0.62, 3E-4)),
                          Gender=rep(c("Female", "Male", "Male", "Female"), each=3))

# This is how the df looks like:
head(mydf)
#> # A tibble: 6 x 4
#>   Subject Magnitude      Value Gender
#>   <chr>   <chr>          <dbl> <chr> 
#> 1 Alice   SomeValue  170       Female
#> 2 Alice   Percent      0.6     Female
#> 3 Alice   Scientific   0.00027 Female
#> 4 Bob     SomeValue  180       Male  
#> 5 Bob     Percent      0.8     Male  
#> 6 Bob     Scientific   0.00025 Male
#  Subject  Magnitude   Value Gender
#1   Alice  SomeValue 1.7e+02 Female
#2   Alice    Percent 6.0e-01 Female
#3   Alice Scientific 2.7e-04 Female
#4     Bob  SomeValue 1.8e+02   Male
#5     Bob    Percent 8.0e-01   Male
#6     Bob Scientific 2.5e-04   Male

scales_y <- list(
  Percent = scale_y_continuous(labels=percent_format()),
  SomeValue = scale_y_continuous(),
  Scientific = scale_y_continuous(labels=scientific_format())
)

# Here the percent (top) is well written, the other ones don't make sense
ggplot(mydf) +
  geom_point(aes(x=Subject, y=Value, color=Gender)) +
  facet_grid(Magnitude~., scales = list(y = scales_y))

Created on 2018-07-10 by the reprex package (v0.2.0).

Merge vs not merge reasons:

• Merge:
  • The diff is small.
  • Backwards compatibility is preserved.
• Not merge:
  • This feature could fit in another package, providing a custom facet. There would be some code duplication, but it looks easy to do (I have a proof of concept).

[clauswilke]

I have a number of questions:

1. Should this be implemented for facet_wrap() as well?

2. What happens with coord_fixed()?

3. What happens when people try to set individual titles for the different scales?

4. When I have a grid of plots with multiple columns and rows, can I set a different scale for each plot?

5. What happens when facets are renamed via a labeller function?

6. What happens when people define a scale outside of the facet_grid() call, and does it matter whether it's called before or after facet_grid()?

This may be a case where implementing this first in a package to gain some experience with this feature and explore the best interface may be a good idea.

[zeehio]

Thanks @clauswilke . Following your advice I have created https://github.com/zeehio/facetscales and I have opened an issue with your comments (zeehio/facetscales#1)

I will try to address those issues (and other issues to come) and get back to you with an improved proposal. For now, I will close the pull request and redirect people from #1613 to my package. Let's see what comes out from this. 👍

Thanks for your feedback!

[lock]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/