Aggregate percentage of total in bar plot-CodePudding

I have a dataset like this (sample)

data.frame(loan_status=c('Y','N','Y','N','Y','N'), property_area=c('Semiurban', 'Urban', 'Rural', 'Semiurban', 'Urban', 'Rural'))

On which loan_status=='Y' means the loan was accepted.

I want to plot the acceptance percentage of each property_area, something like this:

I tried to use ggplot library without success..

What's the easiest way?

CodePudding user response：

You can start with something like that. The scales library allows to easily transform labels to percentages.

library(scales); library(ggplot2); library(dplyr)

df %>% group_by(property_area) %>% 
  summarise(acceptance_pct = sum(loan_status == 'Y') / n()) %>% 
  ggplot(aes(x = acceptance_pct, y = property_area))  
  geom_col(fill = "darkgreen")  
  geom_text(aes(label = percent(acceptance_pct)), hjust = -0.1)  
  scale_x_continuous(labels = percent_format(accuracy = 1))

CodePudding user response：

Here's a base R solution including changing hue based on percentage. I extend your data to show different percentages, not just 50%. Also, I changed "Y" to TRUE and "N" to FALSE.

# adjust data
df1$loan_status[df1$loan_status == "Y"] <- T
df1$loan_status[df1$loan_status == "N"] <- F

# get groups
pl <- aggregate( loan_status ~ property_area, df1, function(x) 
  sum(as.logical(x))/length(x) )

# adjust plotting area
par(mar=c(14,6,14,4))
# plot data, add percentages and grid
bp <- barplot( pl$loan_status, 
               names=pl$property_area, 
               horiz=T, border=F, las=1, 
               col=rgb(0.1, 1-pl$loan_status, 0.1, 1), 
               xlim=c(0,1.14) );
text(pl$loan_status, bp, labels=paste(round(pl$loan_status*100, digits=2),"%"), pos=4);
abline(v = seq(0,1,by=0.2), col = "grey", lty = "dotted")

Data

df1 <- structure(list(loan_status = c("Y", "N", "Y", "N", "Y", "N", 
"Y", "Y", "Y", "N"), property_area = c("Semiurban", "Urban", 
"Rural", "Semiurban", "Urban", "Rural", "Semiurban", "Semiurban", 
"Semiurban", "Rural")), row.names = c("1", "2", "3", "4", "5", 
"6", "7", "8", "9", "10"), class = "data.frame")

CodePudding user response：

Could you start with this?

p = ggplot(mydataf, aes(y = foo))   
    geom_bar(aes(x = (..count..)/sum(..count..)))   
    scale_x_continuous(formatter = 'percent')