Plotting prevalence within bins, denominator to be bin-total rather than entire total

[CM1]

Hi,

I'm trying to plot prevalence for a group of age-ranges and can do so but prevalence for each bin (age_group) is as % of entire population rather than for the number of people within the bin. Any ideas how to amend code below, is it somewhere in the summarise or mutate(pct lines?

Thanks

trimmed_df %>%
group_by(age_group, MM_binary) %>%
summarise(n = n(age_group)) %>%
mutate(pct = round(n/sum(n)*100, 1)) %>%
ggplot(aes(x = age_group, y = pct)) +
geom_col() +
scale_y_continuous(limits = c(0, 70),
breaks = scales::pretty_breaks(),
labels = scales::percent_format(scale = 1)) +
theme(legend.position = "top") +
labs(title = "Prevalence of MM by age",
y = "% of population",
x = "Age group", color = "grey20", size = 20, angle = 90) -> figure_1

figure_1

[nirgrahamuk]

@CM1

summarise(n = n(age_group))

I got a straight syntax error for this, but I think the age_group mention is superfluous ? excluding it seemed to show me what I expect you are attempting.

in the first part, I show the setup that I estimate that you have.
in the second part, are changes I would make

library(tidyverse)

set.seed(42)

trimmed_df <- data.frame(
  age_group=sample(letters[1:5],100,replace=TRUE),
  MM_binary=sample(c(0,1),100,replace=TRUE)
)

your_summary <- trimmed_df %>%
  group_by(age_group, MM_binary) %>%
  summarise(n = n()) %>%
  mutate(pct = round(n/sum(n)*100, 1)) 

your_summary %>%
  mutate(pct = round(n/sum(n)*100, 1)) %>%
  ggplot(aes(x = age_group, y = pct)) +
  geom_col() +
  scale_y_continuous(limits = c(0, 70),
                     breaks = scales::pretty_breaks(),
                     labels = scales::percent_format(scale = 1)) +
  theme(legend.position = "top") +
  labs(title = "Prevalence of MM by age",
       y = "% of population",
       x = "Age group", color = "grey20", size = 20, angle = 90) -> figure_1

figure_1

alt_summary <-  trimmed_df %>%
  group_by(age_group, MM_binary) %>%
  summarise(n = n()) %>% pivot_wider(id_cols="age_group",
                                     names_from="MM_binary",
                                     values_from="n") %>% 
  mutate(frac=`1`/(`1`+`0`),
         labels=scales::percent(frac))

alt_summary

alt_summary %>%
  ggplot(aes(x = age_group, y = frac)) +
  geom_col() +
  scale_y_continuous(limits = c(0, 1),
                     breaks = scales::pretty_breaks(),
                     labels = scales::percent_format(scale = 100)) +
  theme(legend.position = "top") +
  labs(title = "Prevalence of MM by age",
       y = "% of population",
       x = "Age group", color = "grey20", size = 20, angle = 90) -> figure_2

figure_2

[CM1]

I am so unbelivably grateful, this worked first time. I have been battling with this for week, thank you!!!

[CM1]

Just another question about this - I have 6 disease measures altogether (MM_binary and then 5 other ones). Can this plot be made into a line graph where each line represents prevalence for each disease?

Thanks again,

Clare