Fixing elongated figures next to each other in R - r

Question

I'm trying to replicate the theme of these graph using ggplot, I searched online for articles and question to show me how to assign these plots the right size and position and also to assign the tight dot shape, and I found few articles that discussed changing position, I tried the following:
d1<-read.csv("./data/games.csv")
p.1<-ggplot(d1, aes(x=cream_rating, y=charcoal_rating)) +
# Map winner on color. Add some transparency in case of overplotting
geom_point(aes(color = winner), alpha = 0.2) +
# Add the cross: Add geom_pints with one variable fixed on its mean
geom_point(aes(x = mean(cream_rating), color = winner), alpha = 0.2) +
geom_point(aes(y = mean(charcoal_rating), color = winner), alpha = 0.2) +
scale_shape_manual(values=c(16, 17)) +
# "draw"s should be dropped and removed from the title
scale_color_manual(values = c(cream = "seagreen4", charcoal = "chocolate3", draw = NA)) +
ggtitle("Rating of Cream vs Charcoal") +
xlab("rating of cream") + ylab("rating of charcoal") + theme_classic() + theme(plot.title = element_text(hjust = 0.5))
I tried the following to place them together:
require(gridExtra)
plot.1<-p.1
plot.2<-ggExtra::ggMarginal(p.1, type = "histogram")
grid.arrange(plot.1, plot.2, ncol=3)
library(cowplot)
theme_set(theme_cowplot())
plot.1<-p.1
plot.2<-ggExtra::ggMarginal(p.1, type = "histogram")
plot_grid(plot.1, plot.2, labels = "AUTO")
cowplot::plot_grid(plot.1, plot.2, labels = "AUTO")
library(magrittr)
library(multipanelfigure)
figure1 <- multi_panel_figure(columns = 2, rows = 1, panel_label_type = "none")
# show the layout
figure1
figure1 %<>%
fill_panel(plot.1, column = 1, row = 1) %<>%
fill_panel(plot.2, column = 2, row = 1) %<>%
figure1
This is my data set structure:
structure(list(rated = c(FALSE, TRUE, TRUE, TRUE, TRUE, FALSE,
TRUE, FALSE, TRUE, TRUE), turns = c(13L, 16L, 61L, 61L, 95L,
5L, 33L, 9L, 66L, 119L), victory_status = structure(c(3L, 4L,
2L, 2L, 2L, 1L, 4L, 4L, 4L, 2L), .Label = c("draw", "mate", "outoftime",
"resign"), class = "factor"), winner = structure(c(2L, 1L, 2L,
2L, 2L, 3L, 2L, 1L, 1L, 2L), .Label = c("charcoal", "cream",
"draw"), class = "factor"), increment_code = structure(c(3L,
7L, 7L, 5L, 6L, 1L, 1L, 4L, 2L, 1L), .Label = c("10+0", "15+0",
"15+2", "15+30", "20+0", "30+3", "5+10"), class = "factor"),
cream_rating = c(1500L, 1322L, 1496L, 1439L, 1523L, 1250L,
1520L, 1413L, 1439L, 1381L), charcoal_rating = c(1191L, 1261L,
1500L, 1454L, 1469L, 1002L, 1423L, 2108L, 1392L, 1209L)), row.names = c(NA,
10L), class = "data.frame")
This is what I want to achieve:
I tried Stefan's suggestion (which was great help) with some modifications:
d1<-read.csv("./data/games.csv")
ggplot(d1, aes(x=cream_rating, y=charcoal_rating)) +
##### Map winner on color. Add some transparency in case of overplotting
geom_point(aes(color = winner), alpha = 0.2) +
##### Add the cross: Add geom_pints with one variable fixed on its mean
geom_point(aes(x = mean(cream_rating), color = winner), alpha = 0.2) +
geom_point(aes(y = mean(charcoal_rating), color = winner), alpha = 0.2) +
scale_shape_manual(values=c(16, 17)) +
##### "draw"s should be dropped and removed from the title
scale_color_manual(values = c(cream = "seagreen4", charcoal = "chocolate3", draw = NA)) +
ggtitle("Rating of Cream vs Charcoal") +
xlab("rating of cream") + ylab("rating of charcoal") + theme_bw() + theme(plot.title = element_text(hjust = 0.5))
I want to filter out "draw" from the plot, also when I change the dot shapes to triangles and circle, they don't seem to be changing, in addition I get this error:
Warning message:
“Removed 950 rows containing missing values (geom_point).”
Warning message:
“Removed 950 rows containing missing values (geom_point).”
Warning message:
“Removed 950 rows containing missing values (geom_point).”
One more thing that I noticed, I get double cross instead of one!
This is my output:
When I try the first code block in this question, I get long distorted figures not square next to each other.

Answer 1

Maybe this fits your need. To glue the three plots together I make use of the cowplot package. The legend is probably still not perfect.
To get only one legend but still a nice alignment of the plots I made the legends for the first and the third plot "transparent" vis guide_legend and theme options
To make all plots the same size I added transparent marginals to the scatter plot
To fix the position and make the plots square I set the same limits for both axes via xlim and ylim and set the aspect ratio to 1 using theme()
library(ggplot2)
library(dplyr)
library(cowplot)
# Add a second draw to the example data to make the density work
d1 <- d1 %>%
add_row(winner = "draw", cream_rating = 1002, charcoal_rating = 1250)
# Get the limits
lims <- c(floor(min(d1$cream_rating, d1$charcoal_rating)), ceiling(max(d1$cream_rating, d1$charcoal_rating)))
p1 <- d1 %>%
ggplot(aes(x=cream_rating, y=charcoal_rating, color = winner, shape = winner)) +
geom_point(alpha = 0.2, na.rm = TRUE) +
scale_color_manual(values = c(cream = "seagreen4", charcoal = "chocolate3", draw = "blue")) +
scale_shape_manual(values = c(cream = 16, charcoal = 17, draw = 15)) +
xlim(lims) +
ylim(lims) +
labs(x = "rating of cream", y = "rating of charcoal") +
theme_classic() +
theme(plot.title = element_text(hjust = 0.5), legend.position = "bottom") +
theme(aspect.ratio = 1)
p1_1 <- p1 +
guides(color = guide_legend(override.aes = list(color = c(NA, NA, NA)))) +
theme(legend.text = element_blank(), legend.title = element_blank())
p1_1 <- ggExtra::ggMarginal(p1_1, type = "histogram",
margins = 'both',
size = 5,
position = "identity",
color = NA,
fill= NA)
p2 <- ggExtra::ggMarginal(p1, type = "histogram",
margins = 'both',
size = 5,
groupColour = TRUE,
groupFill = TRUE,
position = "identity"
)
# Make legend transparent
p1 <- p1 +
guides(color = guide_legend(override.aes = list(color = c(NA, NA, NA)))) +
theme(legend.text = element_blank(), legend.title = element_blank())
p3 <- d1 %>%
ggplot(aes(x=cream_rating, y=charcoal_rating, color = winner, shape = winner)) +
geom_density_2d(na.rm = TRUE) +
geom_point(alpha = 0, show.legend = FALSE) +
scale_color_manual(values = c(cream = "seagreen4", charcoal = "chocolate3", draw = "blue")) +
xlim(lims) +
ylim(lims) +
labs(x = "rating of cream", y = "rating of charcoal") +
theme_classic() +
theme(plot.title = element_text(hjust = 0.5),
legend.position = "bottom") +
theme(aspect.ratio = 1)
# Make legend transparent
p3 <- p3 +
guides(color = guide_legend(override.aes = list(color = c(NA, NA, NA)))) +
theme(legend.text = element_blank(), legend.title = element_blank())
p3 <- ggExtra::ggMarginal(p3, d1, type = "density",
margins = 'both',
size = 5,
groupColour = TRUE,
groupFill = TRUE,
position = "identity"
)
plot_row <- plot_grid(p1_1, p2, p3, nrow = 1)
# now add the title
title <- ggdraw() +
draw_label(
"Rating of Cream vs Charcoal",
fontface = 'bold',
x = 0,
hjust = 0
)
final <- plot_grid(
title, plot_row,
ncol = 1,
# rel_heights values control vertical title margins
rel_heights = c(0.1, 1)
)
final
Note Depending on the width and heigth of your plotting device, fixing the aspect ratio adds some white space at the top and bottom. Depending on your final output you probably have to play a bit around with the width and height (and scale), e.g. using
ggsave("final.png", width = 18, height = 6, units = "cm", scale = 1.5)
gives