Overplotting problem as per https://twitter.com/TrevorABranch/status/1227494765760897024

exercise.R

## load data
df <- read.csv('http://bit.ly/CEU-R-numbers')

## analyze the association between x and y
summary(df)
cor(df)
## ..

## visualize the association between x and y
plot(df)
## ..

generate.R

## download data
t <- tempfile()
download.file(
    'http://phylopic.org/assets/images/submissions/202c2ad3-48a7-471d-87ef-c6d8406640e8.original.png',
    destfile = t, mode = 'wb')

## read data
library(png)
img <- readPNG(t)

## record the dimensions of the image
height <- dim(img)[1]
width  <- dim(img)[2]

## convert to 2d
points <- as.data.frame(matrix(img, height * width))
points$x <- rep(1:height)
points$y <- rep(1:width, each = height)

## filter for gray points (contour)
points <- subset(points, V4 > 0.5 & V4 < 0.8)

## sample polygon
np <- 15 * 1e3
points <- points[sample(nrow(points), np, replace = TRUE), ]

## flip axeses
points <- data.frame(x = points$y, y = -points$x)

## render polygon
ggplot(points, aes(x, y)) + geom_point()

## add noise
nn <- 1e4 * 2
noise <- round(data.frame(
    x = rnorm(nn, mean = width / 2, sd = width / 4),
    y = rnorm(nn, mean = -height / 2, sd = height / 4)
))

## render noise
ggplot(noise, aes(x, y)) + geom_point()

## add noise to polygon
df <- rbind(points[, c('x', 'y')], noise)

## visualize all points
ggplot(df, aes(x, y)) + geom_point()
ggplot(df, aes(x, y)) + geom_point(alpha = 0.05)
ggplot(df, aes(x, y)) + geom_point(size = 0.2, alpha = 0.1)
ggplot(df, aes(x, y)) + geom_hex(binwidth = 25)

plot-1.png

plot-2.png

plot-3.png

plot-4.png

points.csv