CODEWITHSUNDEEP
rggplot2ecdf
user2845095
user2845095

Reputation: 515

Adding more information to an eCDF plot in R

I used the following code to get a eCDF plot:

df %>% group_by(group1, group2) %>%   
summarise(n = length(unique(sessionID))) %>%
ggplot(aes(n)) +    
stat_ecdf(geom = "step") +
scale_x_continuous(n.breaks = 30) +   
theme_classic()

enter image description here

I would like to add quartiles, mean, and median to the plot, somewhat similar to the plot below.

enter image description here

Upvotes: 0

Views: 348

Answers (1)

teunbrand
teunbrand

Reputation: 37903

My suggestion is to just calculate the quantiles outside the plot and use an extra data.frame as input to the annotation layers.

library(ggplot2)

# Dummy data
df <- data.frame(n = rpois(100, lambda = 5))

# Quantiles
q <- c(0.2, 0.5, 0.8)

# data.frame for quantiles
qdf <- data.frame(
  q = factor(q),
  x = c(rep(-Inf, length(q)), rep(quantile(df$n, q), 2)),
  y = c(rep(q, 2), rep(-Inf, length(q)))
)

ggplot(df, aes(n)) +
  stat_ecdf() +
  # Line segments to/from the ecdf line
  geom_path(
    data = qdf,
    aes(x = x, y = y, colour = q),
    linetype = "dotted"
  ) +
  # Labels at x
  geom_text(
    data = subset(qdf, is.finite(x) & is.finite(y)),
    aes(x = x, y = 0.5 * y, label = x, colour = q),
    hjust = -1
  ) +
  # Labels at y
  geom_text(
    data = subset(qdf, is.finite(x) & is.finite(y)),
    aes(x - 0.5 * min(x), y = y, label = y, colour = q),
    vjust = -1
  )

Created on 2022-10-27 by the reprex package (v2.0.0)

Upvotes: 2

Related Questions