Quality Control for Statistical Graphics: The graphicsQC Package for R

Abstract

An important component of quality control for statistical graphics software is the ability not only to test that code runs without errors, but also to test that code produces the right result. The simple way to test for the correct result in graphical output is to test whether a test file differs from a control file; this is effective in determining whether a difference exists. However, the test can be significantly enhanced by also producing a graphical image of any difference; this makes it much easier to determine how and why two files differ. This article describes the graphicsQC package for R, which provides functions for producing and comparing files of graphical output and for generating a report of the results, including images of any differences.