High-volume scatterplot are scatterplot created from large datasets where large means more than $10^6$ points. Such plots can be used for monitoring outliers and for comparing shapes of data clouds created from different datasets. These high-volume scatterplots are best viewed via raster-based devices. In R the devices png() and jpeg() are raster-based. Both devices require R to be compiled against X11-support. On machines without the required X11-libraries, R can only be compiled without X11-support (using --without-x11). Using vector-based devices (such as pdf) for high volume-scatterplots leads to very large files when saving the plots to files.

The aim of this post is to investigate different alternatives to png() and jpeg() for creating high-volume scatter plot. See here for some preliminary results.