Use geom_ribbon (ggplot2) to plot difference between two time series with changing colors

Question

I'm trying to visualize netexports (exports - imports) by plotting both exports and imports and coloring the difference between the two series depending on whether netexports are positive or negative.

Using R:

library(tidyverse)

data <- tribble(
  ~year, ~exports, ~imports,
  #--|--|----
  2003, 3, 3.6,
  2004, 4, 8.5,
  2005, 7, 7,
  2006, 9, 7,
  2007, 15, 9,
  2008, 17, 12,
  2009, 7, 8,
  2010, 8, 7,
  2011, 8, 8,
  2012, 7, 9,
  2013, 2, 11,
  2014, 9, 13,
  2015, 5, 15
)

ggplot(data, aes(x = year)) +
  geom_ribbon(aes(ymin = imports, ymax = exports))

This gets me:

Which already colors in the differences between the series, but doesn't tell me which one is higher.

Next I've tried:

ggplot(data, aes(x = year)) +
  geom_ribbon(aes(ymin = imports, ymax = exports, fill = exports > imports))

Which yields:

But there seems to be something wrong here and I'm not sure what it is.

Answer 1

This gets closer, though it still fails where you don't have a row at the point where the values are equal. Depending on your needs, you could calculate the point between 2008 and 2009 where they cross, and artificially add it to your dataset.

data$minval <- pmin(data$imports, data$exports)

ggplot(data, aes(x = year)) +
  geom_ribbon(aes(ymin = minval, ymax = exports), fill = "blue") +
  geom_ribbon(aes(ymin = minval, ymax = imports), fill = "red")

Answer 2

There is the stat_difference() that makes a ribbon that is filled depending on whether the max is higher than min:

library(ggh4x)

df <- data.frame(
  x = 1:100,
  y = cumsum(rnorm(100)),
  z = cumsum(rnorm(100))
)

ggplot(df, aes(x = x)) +
  stat_difference(aes(ymin = y, ymax = z), alpha = 0.3) +
  geom_line(aes(y = y, colour = "min")) +
  geom_line(aes(y = z, colour = "max"))