Fix order of diamond color factor.

[jrnold]

The current color variable in the diamonds dataset is an ordered factor, but the levels are in decreasing order of color quality rather than increasing order.

library("ggplot2")
class(ggplot2::diamonds)
#> [1] "tbl_df"     "tbl"        "data.frame"
levels(ggplot2::diamonds$color)
#> [1] "D" "E" "F" "G" "H" "I" "J"

^{Created on 2019-02-15 by the reprex package (v0.2.1)}

The factor order of diamonds$coloris currently D < ... < J. However, since "D" is the best grade and "J" the words, the levels should be J < ... < D. See https://en.wikipedia.org/wiki/Diamond_color#Grading_fancy_color_diamonds.

This pull request puts the factor levels in the correct order.

[clauswilke]

This PR has the potential to mess up a lot of existing examples and tutorials, so we should only merge if we're absolutely certain that this change is required. And I'm not entirely convinced. For rankings of quality, you can usually rank from worst to best or from best to worst. For example, if you submit a grant proposal to the National Institutes of Health, it will be ranked on a scale from 1 to 9, with 1 being exceptional and 9 terrible.

The current diamonds dataset ranks color from best to worst, and you're proposing to instead rank from worst to best. This may be an improvement in some cases and a regression in others. For example, the simple dplyr code below which counts the number of diamonds of different colors produces a table in the same order as the color table on Wikipedia.

library(tidyverse)

diamonds %>%
  group_by(color) %>%
  tally() %>%
  arrange(color)
#> # A tibble: 7 x 2
#>   color     n
#>   <ord> <int>
#> 1 D      6775
#> 2 E      9797
#> 3 F      9542
#> 4 G     11292
#> 5 H      8304
#> 6 I      5422
#> 7 J      2808

^{Created on 2019-02-16 by the reprex package (v0.2.1)}

[yutannihilation]

This PR has the potential to mess up a lot of existing examples and tutorials, so we should only merge if we're absolutely certain that this change is required.

Agreed. As I commented on #2962, it seems almost impossible to safely deprecate and update a dataset.

But, from this comment on the original issue (jrnold/r4ds-exercise-solutions#199), I feel this is just a documentation problem.

The levels of the ggplot2::diamonds$color are specified correctly in the documentation, but incorrectly in the variable.

Should line this be from D (best) to J (worst)?

ggplot2/R/data.R

Line 11 in 03bd946

#' \item{color}{diamond colour, from J (worst) to D (best)}

[jrnold]

I think that the argument about breaking existing examples is key. So I'm going to close it.

Changing that line in the documentation may help.

Yes, quality can be ordered from best to worst or worst to best. But the ordering of color is inconsistent with that of cut and clarity. Of these, color is the only one that is ordered from best to worst. I suppose an argument for the ordering of color is that for letter grades there is prior knowledge that they tend to be ordered from best to worst, which is why it is different.

> head(diamonds$color)
[1] E E E I J J
Levels: D < E < F < G < H < I < J
> head(diamonds$clarity)
[1] SI2  SI1  VS1  VS2  SI2  VVS2
Levels: I1 < SI2 < SI1 < VS2 < VS1 < VVS2 < VVS1 < IF
> head(diamonds$cut)
[1] Ideal     Premium   Good      Premium   Good      Very Good
Levels: Fair < Good < Very Good < Premium < Ideal

[clauswilke]

@jrnold Would you be willing to submit a PR that improves the documentation?

[jrnold]

Removed the commit with changes to thediamonds dataset. Added the doc changes suggested by @yutannihilation

[yutannihilation]

Thanks, could you update the docs by running devtools::document()?

[yutannihilation]

Thanks, could you update the docs by running devtools::document()?

No worries, I'll do this.

[yutannihilation]

Thanks!

[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/