order bar plots in facet grid in R

Question

I have a barchart and I want to order the bars and calculate percentage of changes from one bar to another and show on the plot. I used factors and tried different methods and everything on stackoverflow, but it gives me all sorts of errors. I managed to order the legend in terms of colors but the bar colours does not have the same order.

and here is my code:

library(ggplot2)

data$Active_period  <- factor(data$Active_period,levels = c("study_year","last_year", "this_year"))
fct_rev(data$Active_period)

ggplot(data, aes(count, Name, fill = Active_period)) +
geom_col(width = 0.8, position = position_dodge2(width = 0.8, preserve = "single"))+
facet_grid(Time ~ site)+
scale_fill_manual(name = " ",values = c("study_year" ="#5d5c5c", "this_year" ="#4475aa", 
"last_year" = "#333239"),limits = c("study_year","last_year", "this_year"))+
theme(legend.position = "bottom", 
          axis.text = element_text(size = 11), 
          axis.title = element_text(size = 11, face = "bold"),
          strip.text.x = element_text(size = 12),
          strip.text.y = element_text(size = 12),
          plot.title = element_text(size = 14, face = "bold"),
          plot.subtitle = element_text(size = 13)))

Here is my data:

Name | Active_period| site |Time | count
A | Last_year| north| mornings |10
A | Last_year| south| mornings |20
A | Last_year| north| evenings|45
A | Last_year| south| evenings|35
A | this_year| north| mornings |80
A | this_year| south| mornings |60
A | this_year| north| evenings|95
A | this_year| south| evenings|120
A | study_year| north| mornings |100
A | study_year| south| mornings |400
A | study_year| north| evenings|220
A | study_year| south| evenings|32
B | Last_year| north| mornings |10
B | Last_year| south| mornings |45
B | Last_year| north| evenings|25
B | Last_year| south| evenings|20
B | this_year| north| mornings |300
B | this_year| south| mornings |250
B | this_year| north| evenings|140
B | this_year| south| evenings|20
B | study_year| north| mornings |10
B | study_year| south| mornings |20
B | study_year| north| evenings|10
B | study_year| south| evenings|20

Answer 1

Here is a solution.
First, I create a custom theme so that the problem code is simpler. This is not really part of the question and it can be defined before the answer.

theme_so_q77804382 <- function(){ 
  theme_grey(base_size = 10) %+replace%
    theme(
      legend.position = "bottom", 
      axis.text = element_text(size = 11), 
      axis.title = element_text(size = 11, face = "bold"),
      strip.text.x = element_text(size = 12),
      strip.text.y = element_text(size = 12, angle = -90),
      plot.title = element_text(size = 14, face = "bold"),
      plot.subtitle = element_text(size = 13)
    )
}

---

The plot

The plot is a horizontal bars plot of counts by name so instead of defining the y axis as the names axis, the recommended way is to define the independent variable in the x axis as usual and then reverse with coord_flip.

Now the problem.

• first, coerce Active_period to factor;
• then, compute the percentage changes grouped by Name, site and Time;
• now that we have the changes from base level study_year, reverse the factor levels to have the plot as asked.

Then just plot it, with the percentages plotted with geom_text. The text must also be grouped and since fill is not a geom_text aesthetic, use group = Active_period. Play a little with hjust in order to have the minus signs visible whenever the changes are negative.

suppressPackageStartupMessages({
  library(dplyr)
  library(ggplot2)
})

data$Active_period[data$Active_period == "Last_year"] <- "last_year"

data %>%
  mutate(Active_period = factor(Active_period,levels = c("study_year","last_year", "this_year"))) %>%
  group_by(Name, site, Time) %>%
  arrange(Active_period) %>%
  mutate(perc = scales::percent(c(0, diff(count))/dplyr::lag(count, default  = 1))) %>%
  ungroup() %>%
  mutate(Active_period = forcats::fct_rev(Active_period)) %>%
  ggplot(aes(Name, count, fill = Active_period)) +
  geom_col(
    width = 0.8, 
    position = position_dodge2(width = 0.8, preserve = "single")
  ) +
  geom_text(
    position = position_dodge2(width = 0.8, preserve = "single"),
    aes(label = perc, group = Active_period),
    hjust = -0.2
  ) +
  scale_fill_manual(
    name = " ", 
    values = c(study_year ="#5d5c5c", this_year = "#4475aa", last_year = "#333239"),
    limits = c("study_year","last_year", "this_year")
  ) +
  coord_flip() +
  facet_grid(Time ~ site) +
  theme_so_q77804382()

^{Created on 2024-01-12 with reprex v2.0.2}

---

Edit

In order to remove the "0%" from the first bars, change the dplyr pipe to

data %>%
  mutate(Active_period = factor(Active_period,levels = c("study_year","last_year", "this_year"))) %>%
  group_by(Name, site, Time) %>%
  arrange(Active_period) %>%
  mutate(
    perc = scales::percent(c(0, diff(count))/dplyr::lag(count, default  = 1)),
    perc = c("", perc[-1L])
  ) %>%
  ungroup() %>%
  mutate(Active_period = forcats::fct_rev(Active_period)) %>%

The change that removes the number is in the 2nd mutate.

Answer 2

In your data you use Last_year while in your factor levels you use last_year. The levels are case sensitive. Once, fixed (I fixed it in the data) the plot looks as follows:

Data

d <- readr::read_delim(
"Name|Active_period|site|Time|count
A|Last_year|north|mornings|10
A|Last_year|south|mornings|20
A|Last_year|north|evenings|45
A|Last_year|south|evenings|35
A|this_year|north|mornings|80
A|this_year|south|mornings|60
A|this_year|north|evenings|95
A|this_year|south|evenings|120
A|study_year|north|mornings|100
A|study_year|south|mornings|400
A|study_year|north|evenings|220
A|study_year|south|evenings|32
B|Last_year|north|mornings|10
B|Last_year|south|mornings|45
B|Last_year|north|evenings|25
B|Last_year|south|evenings|20
B|this_year|north|mornings|300
B|this_year|south|mornings|250
B|this_year|north|evenings|140
B|this_year|south|evenings|20
B|study_year|north|mornings|10
B|study_year|south|mornings|20
B|study_year|north|evenings|10
B|study_year|south|evenings|20", delim = "|", col_types = "ffffi")

Code

library(dplyr)
library(ggplot2)

## Note: I replace Last_year by last_year in the data
## alternatively you could also adapt your levels
## but since all levels are lower case I felt this is more consistent
d2 <- d %>% 
  mutate(Active_period = factor(Active_period %>% 
                                  stringr::str_to_lower(),
                                levels = c("this_year", "last_year", "study_year")))

d2 %>% 
  ggplot(aes(count, Name, fill = Active_period)) +
  geom_col(width = 0.8, position = position_dodge2(width = 0.8, preserve = "single"))+
  facet_grid(Time ~ site) +
  scale_fill_manual(name = " ",
                    values = c("study_year" = "#5d5c5c", "this_year" = "#4475aa", 
                               "last_year" = "#333239"),
                    limits = c("study_year", "last_year", "this_year")) +
  theme(legend.position = "bottom", 
        axis.text = element_text(size = 11), 
        axis.title = element_text(size = 11, face = "bold"),
        strip.text.x = element_text(size = 12),
        strip.text.y = element_text(size = 12),
        plot.title = element_text(size = 14, face = "bold"),
        plot.subtitle = element_text(size = 13))