Machine Learning - Data visualization with R (II)

This article continues presenting different techniques that can be used to communicate data or information by encoding it in graphs.

Scatterplot Matrix

Scatter plots show many points plotted in a Cartesian plane. Each point represents a set of coordinates. A Scatterplot matrix shows a scatter plot for each pair of variables.

# load required packages.
install.packages("ggplot2");library("ggplot2");
install.packages("ggExtra");library("ggExtra");
install.packages("gclus");library("gclus");
install.packages("car");library("car");
install.packages("hexbin");library("hexbin");
install.packages("latticeExtra");library("latticeExtra");
install.packages("rgl");library("rgl");

# 4 quantitative variables.
pairs(~mpg+disp+drat+wt,data=mtcars, main="Simple Scatterplot Matrix")

# 4 quantitative variables and 1 categorical variable.
scatterplot.matrix(~mpg+disp+drat+wt|cyl, data=mtcars,
  	main="Three Cylinder Options")

# 4 quantitative variables.
dta <- mtcars[c(1,3,5,6)] 
dta.r <- abs(cor(dta)) 
dta.col <- dmat.color(dta.r, cm.colors(10))
dta.o <- order.single(dta.r) 
cpairs(dta, dta.o, panel.colors=dta.col, gap=.5,
  main="Variables Ordered and Colored by Correlation" )

Simple Scatter Plots

# 2 quantitative variables.
ggplot(mtcars, aes(x=wt, y=mpg))+geom_point()

# 3 quantitative variables.
ggplot(mtcars, aes(x=wt, y=mpg, size=cyl)) +
  geom_point(shape=21, fill="red")

# 2 quantitative variables and 1, 2 or 3 categorical variables.
ggplot(mtcars, 
  aes(x=wt, y=mpg, 
    color=as.factor(mtcars$cyl), 
    fill=as.factor(mtcars$cyl), 
    shape=as.factor(mtcars$cyl))) +
  geom_point()

# 3 quantitative variables and 1, 2 or 3 categorical variables.
ggplot(mtcars, 
  aes(x=wt, y=mpg, size=hp,
    color=as.factor(mtcars$cyl), 
    fill=as.factor(mtcars$cyl), 
    shape=as.factor(mtcars$cyl))) +
  geom_point()

More 2D Scatter Plots

# Scatter plots with smooth lines.
ggplot(mtcars, 
  aes(x=wt, y=mpg, 
    color=as.factor(mtcars$cyl), 
    fill=as.factor(mtcars$cyl), 
    shape=as.factor(mtcars$cyl))) +
  geom_point()+
  geom_smooth(method=lm, se=TRUE)+
  geom_rug()

# Scatter plots with normal confidence interval ellipses.
ggplot(mtcars, 
  aes(x=wt, y=mpg, 
    color=as.factor(mtcars$cyl), 
    fill=as.factor(mtcars$cyl), 
    shape=as.factor(mtcars$cyl))) +
  geom_point()+
  stat_ellipse(type="norm",level=0.9)

# Density courves
ggplot(mtcars, aes(x=wt, y=mpg, colour=as.factor(mtcars$cyl))) +
  stat_density2d()

# Density bins
ggplot(mtcars, aes(x=wt, y=mpg)) +
  geom_bin2d(bins=10)

# Density hexagons
ggplot(mtcars, aes(x=wt, y=mpg)) +
  stat_binhex(bins=10)

# Scatter plot with level plot
levelplot(cyl~wt*mpg, mtcars, 
  panel = panel.levelplot.points, cex = 1) +
  layer_(panel.2dsmoother(..., n = 200))

3D Scatter Plots

In 3D scatter plots, each point represents tree coordinates.

# 3 quantitative variables
scatter3d(
  x=iris$Sepal.Length,y=iris$Sepal.Width,z=iris$Petal.Length, 
  surface=FALSE,
  xlab="SL",ylab="SW",zlab="PL")

# 3 quantitative variables and 1 categorical variable
scatter3d(
  x=iris$Sepal.Length,y=iris$Sepal.Width,z=iris$Petal.Length, 
  surface=FALSE,
  xlab="SL",ylab="SW",zlab="PL")

# 3 quantitative variables with tendency plane.
scatter3d(x=iris$Sepal.Length, y=iris$Sepal.Width,z=iris$Petal.Length,
  xlab="SL",ylab="SW",zlab="PL")

# 3 quantitative variables and 1 categorical variable with confidence interval ellipsoid.
scatter3d(x=iris$Sepal.Length, y=iris$Sepal.Width, 
  z=iris$Petal.Length, 
  groups=iris$Species, surface=FALSE, grid=FALSE, ellipsoid=TRUE,
  surface.col=brewer.pal(n=3, name="Set1"),
  xlab="SL",ylab="SW",zlab="PL")

To be continued...

Share your experience and provide feedback!