Scatterplot Matrix

If we want to show the relationships between $n\ge 2$ variables $\{x_i{\}}_{i=1}^n$ using Scatterplots, we need ${\sum }_{i=1}^n{\sum }_{j>i}^{n}1=n(n-1)/2$ scatterplots. If we plot $x_i\sim x_j$ and $x_j\sim x_i$ as two different plots, then we have $n^2-n$ plots. Then adding labels, we have a scatterplot matrix.

• When passing multiple variables to the R built-in function plot, it will create a scatterplot matrix.
• While this is quite useful for personal exploration of a dataset, it is not recommended for presentation purposes. Something called the Hermann grid illusion makes this plot very difficult to examine.

Other implementations

• ggpairs() in GGally
• splom() in lattice