Empirical cumulative distributions: stat_ecdf

Mix.install([
  {:giggly, path: Path.join(__DIR__, "..")},
  {:kino, "~> 0.19"}
  ])

About this notebook

This notebook follows the examples on ggplot2's reference page for stat_ecdf(), in the order the page gives them. Each example shows ggplot2's code first and then Giggly's. An example Giggly cannot draw yet keeps its ggplot2 code and names what it is waiting for, so the page is also a list of what remains.

stat_ecdf() computes the empirical cumulative distribution of x, or of y: at each value, the proportion of the data at or below it, drawn by default as steps rising from 0 to 1. Unlike a histogram or a density it needs no tuning, and it takes continuous and discrete values alike. With weight mapped, each row raises the curve by its share of the total weight rather than by an equal share. Giggly has none of it yet.

Steps

ggplot2's examples draw their data at random, with R's set.seed(1): 100 values from a normal distribution with standard deviation 3, then 100 with standard deviation 10, and gl(2, 100) labels the two halves 1 and 2.

set.seed(1)
df <- data.frame(
  x = c(rnorm(100, 0, 3), rnorm(100, 0, 10)),
  g = gl(2, 100)
)
ggplot(df, aes(x)) +
  stat_ecdf(geom = "step")

Not in Giggly yet: stat_ecdf() and the steps it draws, geom_step().

Without padding

# Don't go to positive/negative infinity
ggplot(df, aes(x)) +
  stat_ecdf(geom = "step", pad = FALSE)

By default the curve runs from 0 at the left edge of the panel to 1 at the right edge; pad = FALSE stops it at the smallest and largest values.

Not in Giggly yet: stat_ecdf() and its pad option, and geom_step().

One curve per group

# Multiple ECDFs
ggplot(df, aes(x, colour = g)) +
  stat_ecdf()

Not in Giggly yet: stat_ecdf().

A weighted ECDF

# Using weighted eCDF
weighted <- data.frame(x = 1:10, weights = c(1:5, 5:1))
plain <- data.frame(x = rep(weighted$x, weighted$weights))

ggplot(plain, aes(x)) +
  stat_ecdf(linewidth = 1) +
  stat_ecdf(
    aes(weight = weights),
    data = weighted, colour = "green"
  )

plain repeats each x of weighted as many times as its weight, so the green weighted curve falls on the thick unweighted one.

Not in Giggly yet: stat_ecdf(), with or without weight.