How can I automatically highlight multiple sections of the x axis in ggplot2?

user_11829797 0
Viewed 372

I have a line plot that tracks counts over time for multiple factors. A mock version of the data I am working with would be:

step   factor   count
1      a        10
1      b        0
1      c        5
2      a        5
2      b        10
2      c        0
... etc.

The counts are influenced by an external event, and for each step I know whether that event is happening or not. This information could either be in a different dataframe or in the same one, it doesn't really matter, and it would look like this:

step   event
1      FALSE
2      FALSE
...
10     TRUE
11     TRUE
...
30     FALSE
... etc.

I am writing this script to automate the plot creation since I will be dealing with lots of data, and while I know I could use geom_rect() to hard-code highlighting rectangles, it is absolutely not something that I could do manually without wasting way too much time, especially considering the event can turn on and off at different steps in different instances.

Is there any way that I can extract the x limits for geom_rect() dynamically from the data and create as many rectangles as the data set needs? Or is this completely hopeless?

line-plot data-visualization ggplot2 r
2 Answers

This may be a bit hacky, but I think it gives the result you are looking for. Let me create some data first that roughly corresponds to yours:

df <- data.frame(step = rep(1:100, 3), group = rep(letters[1:3], each = 100),
                 value = c(cumsum(c(50, runif(99, -1, 1))), 
                           cumsum(c(50, runif(99, -1, 1))),
                           cumsum(c(50, runif(99, -1, 1)))))

df2 <- data.frame(step = 1:100, event = sample(c(TRUE, FALSE), 100, TRUE))

So the starting plot from df would look like this:

ggplot(df, aes(step, value, colour = group)) + geom_line()

enter image description here

and the event data frame looks like this:

head(df2)
#>   step event
#> 1    1 FALSE
#> 2    2 FALSE
#> 3    3 FALSE
#> 4    4  TRUE
#> 5    5 FALSE
#> 6    6  TRUE

The idea is that you add a semi-transparent red geom_area to the plot, making FALSE values way below the bottom of the range and TRUE values way above the top of the range, then just set coord_cartersian so that the y limits are near to the limits of your main data. This will give you red vertical bands whenever your event is TRUE:

ggplot(df, aes(step, value, colour = group)) + 
  geom_line() + 
  geom_area(data = df2, aes(x = step, y = 1000 * event), 
            inherit.aes = FALSE, fill = "red", alpha = 0.2) + 
  coord_cartesian(ylim = c(40, 60)

enter image description here

user_125003150

Here's an alternative approach to @Allan's excellent approach which relies on preprocessing the event data into groups with dplyr:

library(dplyr)
data2 %>% 
  group_by(group = cumsum(c(1,diff(event))!=0)) %>%
  dplyr::filter(event == TRUE & (step == min(step) | step == max(step))) %>%
ggplot() +
  geom_ribbon(aes(x = step, group = group, ymax = Inf, ymin = -Inf), 
              fill = "yellow", alpha = 0.3) +
  geom_line(data = data, aes(x = step, y = count, color = factor)) +
  facet_wrap(.~factor, ncol = 1)

enter image description here

set.seed(3)  
data <- data.frame(step = rep(1:30, each = 3), factor = rep(letters[1:3],times = 30), count = round(runif(90,0,100)))
data2 <- data.frame(step = 1:30, event = rep(c(TRUE,FALSE,TRUE,FALSE,TRUE,FALSE), c(3,7,2,8,4,6)))
data2
   step event
#1     1  TRUE
#2     2  TRUE
#3     3  TRUE
#...
#28   28 FALSE
#29   29 FALSE
#30   30 FALSE
user_130953260
Related