ggpointdensity plot: A ggplot extension combining scatter plot with density plot

Scatterplots are great for visualizing relationship between two quantitative variables. However, it suffers from overplotting when the number of data points is large. Another solution is plot the density instead of actjual points. However, sometimes you might want to see the datapoints.

ggpointdensity is a ggplot2 extension that solves the problem by combining scatterplot with density plot. You can install ggpointdensity package using

install.packages("ggpointdensityplot")

In this post we will see an example of combining scatterplot with density using ggpointdensity package. Let us load the packages needed.

library(ggplot2)
library(dplyr)
library(viridis)
library(ggpointdensity)
theme_set(theme_bw())

Let us simulate some data to make a scatter plot.

df <-  tibble(x = rnorm(5000, mean=0,sd = 1),
              y = rnorm(5000, mean=5,sd = 10))

Let us start with making a simple scatter plot and see the problem of overplotting with large dataset.

df %>% 
  ggplot( mapping = aes(x = x, y = y)) +
  geom_point()

Scatter plot with overlapping data points

One simple solution is to add transparency to see the overlapping datapoints. It can be of help when the data size is not very big.

df %>% 
  ggplot( mapping = aes(x = x, y = y)) +
  geom_point(alpha=0.3)

A better solution is to combine scatterplot with density plot using ggpointdensity. Instead of geom_point() function, we use geom_pointdensity() function from ggpointdensity package. It is a ggplot2 extension as it offers new geom_* function and we can use it with ggplot2 code.

Avoiding overplotting of datapoints with transparency

ggplot(data = df, mapping = aes(x = x, y = y)) +
  geom_pointdensity()

We can also add colors to reflect the density of datapoints using scale_color_virdis() function from viridis color pallete package.

ggplot(data = df, mapping = aes(x = x, y = y)) +
  geom_pointdensity() +
  scale_color_viridis()