To make it clear, I am looking for a simple way of adding a 90-degree-rotated histogram or density plot whose x-axis aligns with the y-axis of the example plot given below.
library(ggplot2)
library(tibble)
x <- seq(100)
y <- rnorm(100)
my_data <- tibble(x = x, y = y)
ggplot(data = my_data, mapping = aes(x = x, y = y)) +
geom_line()
Created on 2019-01-28 by the reprex package (v0.2.1)
You can try using geom_histogram or geom_density, however it's a little bit complicated as you have to rotate axis for them (while keeping original orientation for geom_line). I would use geom_violin (which is a density plot, but mirrored). If you want to get only one sided violin plot you can use custom geom_flat_violin geom. It was first posted by @David Robinson on his gists.
I used this geom in different answer, however I don't think that it's a duplicate as you need to put it at the end of the plot and combine with different geom.
Final code is:
library(ggplot2)
ggplot(data.frame(x = seq(100), y = rnorm(100))) +
geom_flat_violin(aes(100, y), color = "red", fill = "red", alpha = 0.5, width = 10) +
geom_line(aes(x, y))
geom_flat_violin code:
library(dplyr)
"%||%" <- function(a, b) {
if (!is.null(a)) a else b
}
geom_flat_violin <- function(mapping = NULL, data = NULL, stat = "ydensity",
position = "dodge", trim = TRUE, scale = "area",
show.legend = NA, inherit.aes = TRUE, ...) {
layer(
data = data,
mapping = mapping,
stat = stat,
geom = GeomFlatViolin,
position = position,
show.legend = show.legend,
inherit.aes = inherit.aes,
params = list(
trim = trim,
scale = scale,
...
)
)
}
GeomFlatViolin <-
ggproto(
"GeomFlatViolin",
Geom,
setup_data = function(data, params) {
data$width <- data$width %||%
params$width %||% (resolution(data$x, FALSE) * 0.9)
# ymin, ymax, xmin, and xmax define the bounding rectangle for each group
data %>%
dplyr::group_by(.data = ., group) %>%
dplyr::mutate(
.data = .,
ymin = min(y),
ymax = max(y),
xmin = x,
xmax = x + width / 2
)
},
draw_group = function(data, panel_scales, coord)
{
# Find the points for the line to go all the way around
data <- base::transform(data,
xminv = x,
xmaxv = x + violinwidth * (xmax - x))
# Make sure it's sorted properly to draw the outline
newdata <-
base::rbind(
dplyr::arrange(.data = base::transform(data, x = xminv), y),
dplyr::arrange(.data = base::transform(data, x = xmaxv), -y)
)
# Close the polygon: set first and last point the same
# Needed for coord_polar and such
newdata <- rbind(newdata, newdata[1,])
ggplot2:::ggname("geom_flat_violin",
GeomPolygon$draw_panel(newdata, panel_scales, coord))
},
draw_key = draw_key_polygon,
default_aes = ggplot2::aes(
weight = 1,
colour = "grey20",
fill = "white",
size = 0.5,
alpha = NA,
linetype = "solid"
),
required_aes = c("x", "y")
)
I'd try it with either geom_histogram or geom_density, the patchwork library, and dynamically setting limits to match the plots.
Rather than manually setting limits, get the range of y-values, set that as the limits in scale_y_continuous or scale_x_continuous as appropriate, and add some padding with expand_scale. The first plot is the line plot, and the second and third are distribution plots, with the axes flipped. All have the scales set to match.
library(ggplot2)
library(tibble)
library(patchwork)
y_range <- range(my_data$y)
p1 <- ggplot(data = my_data, mapping = aes(x = x, y = y)) +
geom_line() +
scale_y_continuous(limits = y_range, expand = expand_scale(mult = 0.1))
p2_hist <- ggplot(my_data, aes(x = y)) +
geom_histogram(binwidth = 0.2) +
coord_flip() +
scale_x_continuous(limits = y_range, expand = expand_scale(mult = 0.1))
p2_dens <- ggplot(my_data, aes(x = y)) +
geom_density() +
coord_flip() +
scale_x_continuous(limits = y_range, expand = expand_scale(mult = 0.1))
patchwork allows you to simply add plots to each other, then add the plot_layout function where you can customize the layout.
p1 + p2_hist + plot_layout(nrow = 1)
p1 + p2_dens + plot_layout(nrow = 1)
I've generally seen these types of plots where the distribution is shown in a "marginal" plot—that is, setup to be secondary to the main (in this case, line) plot. The ggExtra package has a marginal plot, but it only seems to work where the main plot is a scatterplot.
To do this styling manually, I'm setting theme arguments on each plot inline as I pass them to plot_layout. I took off the axis markings from the histogram so its left side is clean, and shrunk the margins on the sides of the two plots that meet. In plot_layout, I'm scaling the widths so the histogram appears more in the margins of the line chart. The same could be done with the density plot.
(p1 +
theme(plot.margin = margin(r = 0, unit = "pt"))
) +
(p2_hist +
theme(axis.text.y = element_blank(),
axis.ticks.y = element_blank(),
axis.title.y = element_blank(),
plot.margin = margin(l = 0, unit = "pt"))
) +
plot_layout(nrow = 1, widths = c(1, 0.2))
Created on 2019-01-28 by the reprex package (v0.2.1)
You could use egg::ggarrange(). So basically what you want is this:
p <- ggplot(data=my_data, mapping=aes(x=x, y=y)) +
geom_line() + ylim(c(-2, 2))
q <- ggplot(data=my_data, mapping=aes(x=y)) +
geom_histogram(binwidth=.05) + coord_flip() + xlim(c(-2, 2))
egg::ggarrange(p, q, nrow=1)
Result
Data
set.seed(42)
my_data <- data.frame(x=seq(100), rnorm(100))
my_data1 <- count(my_data, vars=c("y"))
p1 <- ggplot(data = my_data, mapping = aes(x = x, y = y)) + geom_line()
p2 <- ggplot(my_data1,aes(x=freq,y=y))+geom_line()+theme(axis.title.y = element_blank(),axis.text.y = element_blank())
grid.draw(cbind(ggplotGrob(p1), ggplotGrob(p2), size = "last"))
