SPLOMs

Multivariate Comparisons using Scatter Plot Matrices

A scatter plots matrix (SPLOM) is a great way of comparing multiple continuous variables. The example in Figure fig-splom-a compares all four continuous variables of Fisher’s iris data set and colors each species according to color. This is a data heavy and inherently redundant (symmetrical matrix) plot that is more suitable as an exploratory visualization rather than an explanatory visualization.

Figure 1: An example of a SPLOM (scatter plot matrix) and a correlation matrix.

Correlation matrix

Add an example of a pur correlation matrix here.

Figure 2: ?(caption)

Figure 3: ?(caption)

Combination SPLOM & Correlation Matrix

A correlation matrix is a logical extension of a SPLOM, eliminating the redundancy in favor of statistical summaries. In the example in Figure fig-splom-c, three visualizations of the distribution of each variable are provided:

  • a histogram
  • an overlaid density plot
  • an underlying rug

A LOESS regression line is plotted over each scatter plot and in the corresponding boxes in the upper-right portion, their correlation values are provided with corresponding asterisks for significance. The large amount of summary information presented in this plot also makes it clear that this visualization is more for exploration than presentation.

Figure 4: An example of a combination SPLOM and correlation matrix.