Interactive visualization of data-driven methods for the exploration of spatiotemporal public health data

Exploratory data analysis (EDA) offers an alternative approach that prioritizes flexibility and speed over the thoroughness and statistical rigor of traditional methods. This allows analysts to gain a broad understanding of a problem before proceeding to deeper analysis, which is especially useful when dealing with unfamiliar domains or when rapid insights are needed, such as for hypothesis generation and health surveillance.<br><br>Effective EDA requires complex information to be presented to the analyst at speed and at scale.

Interactive visualization, which capitalizes on our innate visual processing abilities, is a particularly effective means of achieving this. Designing analytical environments that effectively leverage interactive visualization, however, can be challenging. This is especially true when dealing with spatiotemporal datasets, which are fundamental to public health research.

Combining spatial and temporal dimensions, these datasets are inherently complex and difficult to visualize effectively due to the sheer volume of information they contain. In this thesis, we investigate various ways to explore spatial, temporal, and spatiotemporal data. We focus specifically on using data-driven approaches to extract meaningful patterns from these datasets.

This allows for the visualization of important information without overwhelming the analyst --- a crucial consideration in exploratory environments where speed and simplicity are imperative. We introduce various novel analytical approaches, each paired with a web-tool to facilitate easy access and use for researchers. We hope that this work will further the availability of exploratory methods for spatiotemporal public health data, addressing an important gap in the current research landscape.