Plot the means of multiple columns - r

Question

I want to show different barplots for the years and gender with the mean values of the variables Q1 to Q5, which should look like a density.
I have data that looks like this:
data <- data.frame(userid = c(1,1,1,2,2,2,3,3,3),
year = c(2013,2014,2015,2013,2014,2015,2013,2014,2015),
gender = c(1,1,1,0,0,0,0,0,0),
Q1 = c(3,2,3,1,0,1,2,1,0),
Q2 = c(4,3,4,2,0,2,1,4,3),
Q3 = c(1,2,1,3,5,4,5,4,5),
Q4 = c(1,2,1,2,4,3,2,2,1),
Q5 = c(1,1,1,2,1,0,0,0,1))
My solution was to filter() for year and gender first and then use summarise(),
to get a vector of the means and put this into the barplot() function:
data %>% filter(gender==1,year==2013) %>% select(-userid,-gender,-year) %>% summarise_all(mean) %>%
as.numeric() %>%
barplot()
Instead of doing this for every combination of year and gender,
is there a more elegant way, using ggplot and facet_wrap()?

Answer 1

I may have misunderstood how you want the plot arranged, but if you want to show the mean score answer per year and gender group, you could do facets like this:
library(tidyverse)
data %>%
pivot_longer(starts_with("Q")) %>%
group_by(year, gender, name) %>%
summarize(value = mean(value)) %>%
ggplot(aes(name, value)) +
geom_col(fill = 'deepskyblue4') +
facet_grid(year ~ gender) +
labs(x = 'Question', y = 'Average score') +
theme_minimal(base_size = 16)

Answer 2

Maybe you want something like this with facet_wrap and geom_col where the mean is calculate using rowMeans like this:
library(dplyr)
library(ggplot2)
data %>%
mutate(mean = rowMeans(select(., starts_with("Q")), na.rm = TRUE)) %>%
ggplot(aes(x = year, y = mean, fill = factor(gender))) +
geom_col() +
labs(x = 'Year', y = 'Mean Q1 to Q5', fill = 'Gender') +
theme_bw() +
facet_wrap(~userid)
Created on 2022-10-28 with reprex v2.0.2

Answer 3

First, pivot your data from wide to long format, group by year, gender, and Q, and summarize to mean values.
library(tidyr)
library(dplyr)
library(ggplot2)
data_long <- data %>%
pivot_longer(Q1:Q5, names_to = "Q", values_to = "value") %>%
group_by(year, gender, Q) %>%
summarize(value = mean(value), .groups = "drop")
data_long
# A tibble: 30 × 4
year gender Q value
<dbl> <dbl> <chr> <dbl>
1 2013 0 Q1 1.5
2 2013 0 Q2 1.5
3 2013 0 Q3 4
4 2013 0 Q4 2
5 2013 0 Q5 1
6 2013 1 Q1 3
7 2013 1 Q2 4
8 2013 1 Q3 1
9 2013 1 Q4 1
10 2013 1 Q5 1
# … with 20 more rows
Then plot using ggplot2::facet_grid().
ggplot(data_long, aes(Q, value)) +
geom_col() +
facet_grid(year ~ gender)

Answer 4

aggregate then barplot.
par(mfrow=c(1, 4))
sapply(unique(data$year), \(x) {
as.matrix(aggregate(cbind(Q1, Q2, Q3, Q4, Q5) ~ gender, data[data$year == x, ], FUN=mean)[-1]) |>
barplot(beside=TRUE, col=c(2, 4), main=x)
})
plot.new()
legend('left', legend=c('m', 'f'), col=c(2, 4), cex=1.2, pch=15, bty='n')

Answer 5

This approach does not require you to first calculate the mean, that is handled by stat_summary(), specifying fun = mean.
library(tidyverse)
data <- data.frame(userid = c(1,1,1,2,2,2,3,3,3),
year = c(2013,2014,2015,2013,2014,2015,2013,2014,2015),
gender = c(1,1,1,0,0,0,0,0,0),
Q1 = c(3,2,3,1,0,1,2,1,0),
Q2 = c(4,3,4,2,0,2,1,4,3),
Q3 = c(1,2,1,3,5,4,5,4,5),
Q4 = c(1,2,1,2,4,3,2,2,1),
Q5 = c(1,1,1,2,1,0,0,0,1))
data %>%
select(starts_with("Q")) %>% # Selects each column that starts with "Q"
pivot_longer(cols = everything()) %>% # Pivot to long format
ggplot(aes(x = name, y = value, fill = name)) +
stat_summary(geom = "bar", fun = "mean") + # Geom and function can be changed easily
theme_classic() +
labs(x = "Q", y = "Mean value", fill = NULL)
Created on 2022-10-28 by the reprex package (v2.0.1)