Effective Use of Small Multiples and Interactivity: Moving Beyond Static Summary Plots

Oregon Convention Center 

Dimension reduction techniques, such as principal component analysis, have been a staple of multivariate exploratory data analysis, modeling, and visualization for more than 100 years. The size and complexity of data are often the main factors that pose challenges to effective exploration. This challenge has been exacerbated in recent decades, as the cost to store data has consistently decreased, and in most fields of study instrumentation and data collection become increasingly cost efficient. In response, more complex and sophisticated dimension techniques, such as uniform manifold approximation and project, have been introduced by the research community. While these techniques provide value, they often lack interpretability and transparency. Here, we introduce several visualization tools to allow decision-makers and data consumers to investigate large datasets in detail. We demonstrate applications of these tools, focusing on the use of Tufte's small multiples concept, coupling these with metrics to aid in data exploration, and linked interactive plots. We will further discuss practical mechanisms for sharing and disseminating data visualizations generated by these tools.