I am trying to reproduce the following graph in R with GGDAG.
So far I am able to reproduce all nodes and arrows. However I do not know how to change the color of the "U" node and add the correspondent legend to the graph. Help is highly appreciated!See code for the uncolored DAG below.
library(ggplot2)
library(ggdag)
coord_dag <- list(
x = c(P = 0, X = 2, D = 2, Z = 2, U = 4, M = 4, Y = 6),
y = c(P = 2, X = 0, D = 2, Z = 6, U = 0, M = 3, Y = 2))
my_dag <- ggdag::dagify(X ~ P,
Z ~ P,
D ~ X Z,
M ~ D Z,
U ~ X,
Y ~ U D M Z,
coords = coord_dag,
exposure = "D",
outcome = "Y")
ggdag::ggdag(my_dag)
theme_dag()
CodePudding user response:
You can use this code:
library(ggplot2)
library(ggdag)
library(dplyr)
coord_dag <- list(
x = c(P = 0, X = 2, D = 2, Z = 2, U = 4, M = 4, Y = 6),
y = c(P = 2, X = 0, D = 2, Z = 6, U = 0, M = 3, Y = 2))
my_dag <- ggdag::dagify(X ~ P,
Z ~ P,
D ~ X Z,
M ~ D Z,
U ~ X,
Y ~ U D M Z,
coords = coord_dag,
exposure = "D",
outcome = "Y") %>%
tidy_dagitty() %>%
mutate(colour = ifelse(name == "U", "Unobserved", "Observed"))
my_dag %>%
ggplot(aes(x = x, y = y, xend = xend, yend = yend))
geom_dag_point(aes(colour = colour))
geom_dag_edges()
geom_dag_text()
theme_dag()
Output:
CodePudding user response:
This is a little tricky since there doesn't seem to be the option directly within the function to label colours how you would like. Perhaps the easiest way would be to add an aesthetic mapping to the point layer:
p <- ggdag::ggdag(my_dag) theme_dag()
p$layers[[3]]$mapping <-
aes(colour = c("Observed", "Unobserved")[as.numeric(name == "U") 1])
p scale_color_manual(values = c("black", "#cc2055"))
theme(legend.position = c(0.8, 0.8))
CodePudding user response:
To change the colour of your y node, add a new column to the tidy DAG and plot with the colour aesthetic:
Sample code:
library(ggplot2)
library(ggdag)
library(dplyr)
coord_dag <- list(
x = c(P = 0, X = 2, D = 2, Z = 2, U = 4, M = 4, Y = 6),
y = c(P = 2, X = 0, D = 2, Z = 6, U = 0, M = 3, Y = 2))
my_dag <- ggdag::dagify(X ~ P,
Z ~ P,
D ~ X Z,
M ~ D Z,
U ~ X,
Y ~ U D M Z,
coords = coord_dag,
exposure = "D",
outcome = "Y") %>%
tidy_dagitty() %>%
dplyr::mutate(colour = ifelse(name == "U", "Unobserved", "Observed"))
ggdag::ggdag(my_dag)
geom_dag_point(aes(colour = colour))
geom_dag_edges()
geom_dag_text()
theme_dag()
Plot: