ggplot: Drawing tiles / rectangles with discrete variables - r

Question

I'm attempting to draw tiles / rectangles to get the following result:
library(tidyverse)
library(plotly)
set.seed(0)
df <- tibble(
a = runif(5),
b = runif(5),
c = runif(5),
d = runif(5),
case_id = 1:5
) %>% tidyr::pivot_longer(cols = -case_id)
plot <- ggplot2::ggplot(
data = df,
mapping = aes(
x = name,
y = value,
group = case_id
)
) + geom_point()
plot_boxes_y <- seq(from = 0, to = 1, by = .2)
plot_boxes_x <- unique(df$name) %>% length()
for (x in 1:plot_boxes_x) {
for (y in plot_boxes_y) {
plot <- plot + geom_rect(
mapping = aes_(
xmin = x - .5,
xmax = x + .5,
ymin = y - .5,
ymax = y + .5
),
color = "red",
fill = NA
)
}
}
plotly::ggplotly(plot)
As you can see, I currently do this by looping through coordinates and drawing each rectangle individually. The problem is, that this generates many layers which makes plotly::ggplotly() really slow on large datasets.
Therefore, I'm looking for a more efficient way. Please note, that I cannot use the panel.grid, since I intend to visualize z-data by filling rectangles later on.
My approach was to draw geom_tile() on top of the scatter plot:
# my attempt
df$z <- rep(0, nrow(df))
plot2 <- ggplot2::ggplot(
data = df,
mapping = aes(
x = name,
y = value,
color = z,
group = case_id
)
) + geom_point() + geom_tile()
I assume that this fails because of the fact that name is a discrete variable? So, how can i efficiently draw tiles in addition to my scatterplot?
Thanks

Answer 1

Here is a solution using the geom_tile option. The key here creating a data frame to hold the coordinates of the grid and then specifying the aesthetics individually in each of the function calls.
library(ggplot2)
library(tidyr)
set.seed(0)
df <- tibble(
a = runif(5),
b = runif(5),
c = runif(5),
d = runif(5),
case_id = 1:5
) %>% pivot_longer(cols = -case_id)
df$z <- rep(0, nrow(df))
#make data frame for the grid corrdinates
grid<-data.frame(x=factor( ordered( 1:4), labels = c("a", "b", "c", "d" )),
y=rep(seq(0, 1, .1), each=4))
#plot using geom_tile & geom_point
plot2 <- ggplot2::ggplot() + geom_tile(data=grid, aes(x=x, y=y), fill=NA, col="red") +
geom_point(data = df,
mapping = aes(
x = name,
y = value,
color = z,
group = case_id))
print(plot2)

Answer 2

if you don't mind them going beyond the axis
ggplot(df,aes(x=name,y=value)) + geom_point() +
geom_vline(xintercept=seq(0.5,4.5,by=1)) +
geom_hline(yintercept=seq(0,2,by=.2))
else:
#make a new data frame
GRIDS = rbind(
# the vertical lines
data.frame(x=seq(0.5,4.5,by=1),xend=seq(0.5,4.5,by=1),y=0,yend=2),
# the horizontal lines
data.frame(x=0.5,xend=4.5,y=seq(0,2,by=.2),yend=seq(0,2,by=.2))
)
ggplot(df,aes(x=name,y=value)) + geom_point() +
geom_segment(data=GRIDS,aes(x=x,y=y,xend=xend,yend=yend),col="red")