I would like to define a custom line type for a graph in R.
As a minimal working example, consider the following
plot(1:5,runif(5),type="l",lty="longdash")
which gives me just some random line (like the dashed black line in the picture below). Is there a way to instead make the dashes look like small arrows (like in the red example, which I constructed with illustrator)?
I continued on the work of @cardinal40. The added value is that in my solution, there is space between the consecutive arrows.
df2 %>%
left_join(df, by = "x") %>%
mutate(
y = if_else(is.na(y), approx(x, y, x)$y, y),
## calculate the change in x and y, and take 20%
x_delta = abs(x-lead(x))*.2,
y_delta = abs(y -lead(y))*.2,
### provide the relative space calculate above but ...
### treat differently depending on descending or ascending
xend = ifelse(x > lead(x), lead(x)+x_delta, lead(x)-x_delta),
yend = ifelse(y > lead(y), lead(y)+y_delta, lead(y)-y_delta)
) %>%
filter(!is.na(xend)) %>%
ggplot(aes(x = x, y = y, xend = xend, yend = yend)) +
geom_segment(arrow = arrow(length = unit(0.1, "inches")))
This yields the following graph:
library(tidyverse)
set.seed(1)
# Original data
df <- tibble(x = 1:5, y = runif(5))
# Intermediate points
df2 <- tibble(x = seq(min(df$x), max(df$x), by = 0.25))
# Join, interpolate and plot
df2 %>%
left_join(df, by = "x") %>%
mutate(
y = if_else(is.na(y), approx(x, y, x)$y, y),
xend = lead(x),
yend = lead(y)
) %>%
filter(!is.na(xend)) %>%
ggplot(aes(x = x, y = y, xend = xend, yend = yend)) +
geom_segment(arrow = arrow(length = unit(0.1, "inches")))
The gaps between the arrows are a complicating factor but this overall setup could be a good starting point.
