rotation direction set up randomly for pie chart.

[withr]

I found a problem with geom_bar

I expected the brown part of the second pie chart should be on the left, but it didn't.

I found it depends on the order of color code. How to fix it?

Here is the code to reproduce the bug:

dat1 <- data.frame(year = c(2012, 2019), 
                   value = c(1, 1), 
                   color = c("#1CF900", "#FFA600"))
dat2 <- data.frame(year = c(2019, 2023), 
                   value = c(1, 1), 
                   color = c("#FFA600", "#FF4000"))

pie1 <- ggplot(dat1, aes(x = "", y = value, fill = color)) +
  geom_bar(stat = "identity", width = 1) +
  coord_polar(theta = "y", start = 0) +
  theme_void() +
  theme(legend.position = "none") +
  scale_fill_identity() 

pie2 <- ggplot(dat2, aes(x = "", y = value, fill = color)) +
  geom_bar(stat = "identity", width = 1) +
  coord_polar(theta = "y", start = 0) +
  theme_void() +
  theme(legend.position = "none") +
  scale_fill_identity() 

print(pie1 + pie2)

[clauswilke]

The ordering of categorical variables is determined by the data values (alphabetical by default), and since you're using color codes as data values they are used for ordering.

I would strongly advise against using scale_fill_identity() unless you truly know what you're doing.

Also, this may help: https://wilkelab.org/SDS375/slides/getting-things-in-order.html#1

[withr]

Thanks for your quick reply! But, the data value are 1 for both categories, and I didn't use color codes as data values. the color codes is for color only. Even I changed the value, the output didn't change. Please take a further look of this issue. @clauswilke

[teunbrand]

I agree with Claus that this is not a bug. These issues here on github are for tracking bugs and feature requests. The issue you've surfaced here is a usage question which would have a better home on stackoverflow or the posit community.

[withr]

@teunbrand Can you provide a script to solve the problem if you take this as a usage question?

[teunbrand]

This is not the place for that. You've already been directed to more resources by Claus: https://wilkelab.org/SDS375/slides/getting-things-in-order.html#1. If these are insufficient to resolve your problem, I'll repeat that a better venue for usage questions is stackoverflow or the posit community.

[withr]

Sad that none of you take this as bug. If the function can't generate consistent output, it's a bug.

[smouksassi]

it is unclear what are you trying to achieve and what is expected.

library(ggplot2)
library(patchwork)
dat1 <- data.frame(year = c("2012", "2019"), 
                   value = c(1, 1), 
                   pie = "one")
dat2 <- data.frame(year = c("2019", "2023"), 
                   value = c(1, 1), 
                   pie="two")
dat <- rbind(dat1,dat2)
ggplot(dat, aes(x = pie, y = value, fill = year)) +
  geom_bar(stat = "identity", width = 1) +
  facet_grid(~pie,scales="free")+
  scale_fill_manual(values=c("#1CF900","#FFA600", "#FF4000"))

^{Created on 2024-12-19 with reprex v2.1.1}

[clauswilke]

I didn't use color codes as data values

You did. They're in your data frame. And then you chose to display them, via scale_fill_identity().

As I wrote in my first response, scale_fill_identity() is almost always the wrong choice, and you shouldn't use it unless you're an expert and know exactly what you're doing.