One of the biggest challenges statisticians face when working with non-statisticians on applied problems is to be able to effectively communicate the statistical results. In this paper we discuss the use of interactive visualization as a tool to present the relationship between a binary response and a set of explanatory variables. The visualization system we present allows users to "manipulate" directly, dynamically, and interactively their data set. At a first level, this allows to integrate visualization with a classical statistical analysis by providing interactive 3D views of the data set. Beyond its potential use as a straightforward visualization tool, this new system opens up interesting possibilities for exploring data visually, by its better exploitation of the human visual system. The paper presents an example of exploring visual relationships between environmental variables and the presence/absence of Lyme disease in Rhode Island.
The CAUSE Research Group is supported in part by a member initiative grant from the American Statistical Association’s Section on Statistics and Data Science Education