Reputation: 446
I have been trying to output in ggplot a graph that shows labels in percentage value and in proportion to the grouping factor defined in geom_bar. Instead of % values proportionate to the overall population, I would like to output a label value that is proportionate to each sub-group (in this case Place A and Place B) but I have not managed to. See below the reproducible example
Reproducible dataframe
Random<-data.frame(replicate(3,sample(0:3,3024,rep=TRUE)))
Random$Trxn_type <- sample(c("Debit", "Credit"),
size = nrow(Random),
prob = c(0.76, 0.24), replace = TRUE)
Random$YN <- sample(c("Yes", "No"),
size = nrow(Random),
prob = c(0.76, 0.24), replace = TRUE)
Random$Place <- sample(c("PlaceA", "PlaceB"),
size = nrow(Random),
prob = c(0.76, 0.24), replace = TRUE)
Random<-Random[, 4:6]
Then applied the following code
Share<-ggplot(Random, aes(x = YN, fill=Place)) +
scale_fill_brewer(palette="Greens")+
geom_bar(aes(y = ..prop.., group = Place),position = position_dodge()) +
facet_wrap(~ Random$Trxn_type, scales = "free_x", ncol=2)+
theme(strip.text.x = element_text(size = 15, colour = "black"))+
theme(panel.background = element_rect(fill = "white"),legend.position = "bottom")+
scale_y_continuous(labels = percent)+
ylab("Frequency") +
coord_flip()+
xlab("Answers") +
theme(plot.title = element_text(size = 16, face = "bold"),
axis.text=element_text(size=12),
axis.title=element_text(size=12))+
geom_text(aes(y=..prop..,label=scales::percent((..count..)/tapply(..count..,..PANEL..,sum)[..PANEL..])),
stat="count", vjust=-.5, position=position_dodge(.9))
Share
And got the following output
Instead of this percentage distribution I would like to see the % value of replies considering Place A and Place B as two separate populations. Put it more simply I would like the labels to show the % value corresponding to the size of the histogram bars in a way that histograms for Place A in credit to sum up to 100 and histograms for Place B in credit to sum up to 100. The same would apply to debit.
Thanks!
Upvotes: 1
Views: 985
Reputation: 76402
Here is a solution that computes the proportions with dplyr
and then pipes the result to ggplot
.
I have also put all theme
settings in the same call to theme()
.
I have reposted the data creation code, this time setting the RNG seed in order to make the data example reproducible.
library(dplyr)
library(ggplot2)
Random %>%
count(Trxn_type, YN, Place) %>%
left_join(Random %>% count(Trxn_type, name = "m"), by = "Trxn_type") %>%
mutate(Prop = n/m) %>%
ggplot(aes(x = YN, y = Prop, fill = Place)) +
geom_col(position = position_dodge()) +
geom_text(aes(label = scales::percent(Prop)),
hjust = -0.25,
position = position_dodge(0.9)) +
facet_wrap(~ Trxn_type, scales = "free_x", ncol = 2) +
scale_fill_brewer(palette = "Greens") +
scale_y_continuous(limits = c(0, 1), labels = scales::percent) +
xlab("Answers") +
ylab("Frequency") +
coord_flip() +
theme(panel.background = element_rect(fill = "white"),
legend.position = "bottom",
strip.text.x = element_text(size = 15, colour = "black"),
plot.title = element_text(size = 16, face = "bold"),
axis.text = element_text(size = 12),
axis.title = element_text(size = 12))
Edit.
Following the OP's comment, here is a way to also count by Place
. The only change to the code above is the left_join
instruction.
left_join(Random %>% count(Trxn_type, Place, name = "m"),
by = c("Trxn_type", "Place")) %>%
Data creation code.
set.seed(1234)
Random <- data.frame(replicate(3,sample(0:3,3024,rep=TRUE)))
Random$Trxn_type <- sample(c("Debit", "Credit"),
size = nrow(Random),
prob = c(0.76, 0.24), replace = TRUE)
Random$YN <- sample(c("Yes", "No"),
size = nrow(Random),
prob = c(0.76, 0.24), replace = TRUE)
Random$Place <- sample(c("PlaceA", "PlaceB"),
size = nrow(Random),
prob = c(0.76, 0.24), replace = TRUE)
Random <- Random[, 4:6]
Upvotes: 1