Side-by-Side plots lined up in R - r

Question

I am trying to place two plots side-by-side in R and have the below example.
library(vioplot)
x <- rnorm(100)
y <- rpois(100,1)
plot(x, y, xlim=c(-5,5), ylim=c(-5,5),type='n')
vioplot(x, col="tomato", horizontal=TRUE, at=-4, add=TRUE,lty=2, rectCol="gray")
vioplot(y, col="cyan", horizontal=TRUE, at=-3, add=TRUE,lty=2)
vioplot(y, col="cyan", horizontal=TRUE, at=-2, add=TRUE,lty=2)
With this data, I'm able to make a vioplot of my x and y variables. Now, for example, I want to develop bar plots of separate count data that relates to each vioplot on the left-hand side.
counts <- c(10, 20, 30)
barplot(counts, main="Car Distribution", horiz=TRUE)
I've used the mtcars example but it could be any count data. I'm wondering if it is possible to generate these plots side-by-side so that the count plot lines up with the vioplot correctly. I do not need any y-axis labels for the count plot.

Answer 1

According your specifications ggplot is my recommendation
library(tidyverse)
p1 <- lst(x, y, y1=y) %>%
bind_cols() %>%
pivot_longer(1:3) %>%
ggplot(aes(name, value)) +
geom_violin(trim = FALSE)+
geom_boxplot(width=0.15) +
coord_flip()
p2 <- mtcars %>%
count(gear) %>%
ggplot(aes(gear, n)) +
geom_col()+
coord_flip()
cowplot::plot_grid(p1, p2)
In base R you can do (please note, I used boxplot, but should work with viopülot either)
par(mfrow=c(1,2))
counts <- table(mtcars$gear)
boxplot(cbind(x,y,y), col="tomato", horizontal=TRUE,lty=2, rectCol="gray")
barplot(counts, main="Car Distribution", horiz=TRUE,
names.arg=c("3 Gears", "4 Gears", "5 Gears"))

Answer 2

Another option if you want to use ggplot is function ggarrange() from ggpubr.
library(dplyr)
library(ggplot2)
library(ggpubr)
# Create a sample dataset
dt <- tibble(group = rep(c("x", "y"), each = 100)) %>%
mutate(value = if_else(group == "x", rnorm(200),
as.double(rpois(200, 1))))
# Combined violin/Box plot
violins <- dt %>%
ggplot(aes(value, group)) +
geom_violin(width = 0.5) +
geom_boxplot(width = 0.1)
# Bar chart
bars <- dt %>%
ggplot(aes(group)) +
geom_bar(width = 0.1) +
coord_flip()
# Combine
ggpubr::ggarrange(violins, bars + rremove("ylab") + rremove("y.text"), ncol = 2)
Output:

Answer 3

You can use this code:
library(vioplot)
x <- rnorm(100)
y <- rpois(100,1)
par(mfrow=c(1,2))
plot(x, y, xlim=c(-5,5), ylim=c(-5,-1),type='n')
vioplot(x, col="tomato", horizontal=TRUE, at=-4, add=TRUE,lty=2, rectCol="gray")
vioplot(y, col="cyan", horizontal=TRUE, at=-3, add=TRUE,lty=2)
vioplot(y, col="cyan", horizontal=TRUE, at=-2, add=TRUE,lty=2)
counts <- table(mtcars$gear)
barplot(counts, main="Car Distribution", horiz=TRUE,
names.arg=c("3 Gears", "4 Gears", "5 Gears"))
Output:

Answer 4

Thank you for your interesting question, which has motivates me to explore base R graphics features. I have tried to find a case where the side-by-side configuration between the violin plot and the barplot provides a meaningful relationship. The case is that I have a subset of iris data with various counts of the species. I want to show three statistics:
the counts of sampled each species, by showing barplots;
the spread of sepal lengths in each sampled species, by showing violin plots; and
the median petal width of each sampled species, by positioning the violin plots.
I follow #GW5's idea here to create barplots of which the positions on the axes can be controlled. I follow #IRTFM's idea here to adjust the origins of the axes.
Here is the full code:
library(vioplot)
some_iris <- iris[c(1:90, 110:139), ]
ir_counts <- some_iris |> with(Species) |> table()
ir_counts
# setosa versicolor virginica
# 50 40 30
ir_names <- names(ir_counts)
ir_colors <- c("cyan", "green", "pink")
x_vio1 <- some_iris |> subset(Species == ir_names[1]) |> with(Sepal.Length)
x_vio2 <- some_iris |> subset(Species == ir_names[2]) |> with(Sepal.Length)
x_vio3 <- some_iris |> subset(Species == ir_names[3]) |> with(Sepal.Length)
y_vio1 <- some_iris |> subset(Species == ir_names[1]) |> with(Petal.Length) |> median()
y_vio2 <- some_iris |> subset(Species == ir_names[2]) |> with(Petal.Length) |> median()
y_vio3 <- some_iris |> subset(Species == ir_names[3]) |> with(Petal.Length) |> median()
# `xpd = FALSE` to keep the grid inside the plotting boxes.
par(mfrow = c(1, 2), xpd = FALSE)
# The violin plots, put on the left side.
plot(NULL,
xlim = c(0, 10), ylim = c(0, 10), type = "n", las = 1, xaxs = "i", yaxs = "i",
xlab = "Sepal Length (cm)", ylab = " Median Petal Width (cm)")
vioplot(x_vio1, col = ir_colors[1], horizontal = TRUE, at = y_vio1, add = TRUE, lty = 2)
vioplot(x_vio2, col = ir_colors[2], horizontal = TRUE, at = y_vio2, add = TRUE, lty = 2)
vioplot(x_vio3, col = ir_colors[3], horizontal = TRUE, at = y_vio3, add = TRUE, lty = 2)
grid()
# The texts that informs the names of the species
text(labels = ir_names, y = c(y_vio1, y_vio2, y_vio3),
x = c (min(x_vio1), min(x_vio2), min(x_vio3)) - 1)
# The barplots, put on the right side.
plot(NULL,
xlim = c(0, 60), ylim = c(0, 10), yaxt = "n", type = "n",
las = 1, xlab = "Counts", ylab = "", xaxs = "i", yaxs = "i"
)
rect(xleft = 0, xright = ir_counts[1],
ybottom = y_vio1 - 0.3, ytop = y_vio1 + 0.3, col = ir_colors[1])
rect(xleft = 0, xright = ir_counts[2],
ybottom = y_vio2 - 0.3, ytop = y_vio2 + 0.3, col = ir_colors[2])
rect(xleft = 0, xright = ir_counts[3],
ybottom = y_vio3 - 0.3, ytop = y_vio3 + 0.3, col = ir_colors[3])
grid()
Here is the result:
In case you want to put labels on the barplots (on the right side), you can use mtext as follows:
# ... (The same code above)
mtext(text = ir_names, side = 2, at = c(y_vio1, y_vio2, y_vio3),
line = 0.2, las = 1 )
The resulted labels: