Visualization Tool: Exploring COVID-19 Data

TL;DR

This paper presents a novel COVID-19 data visualization tool that integrates traditional charts with a Bayesian surprise-based map to enhance understanding of pandemic patterns and avoid common visualization biases.

Contribution

The study introduces a comprehensive COVID-19 visualization platform incorporating a Bayesian surprise map to improve interpretability and reduce misleading visual prominence in spatial data.

Findings

The tool effectively visualizes COVID-19 data from multiple sources.

The Surprising Map reduces bias caused by base rates and sample size artifacts.

Enhanced understanding of COVID-19 spatial patterns.

Abstract

The ability to effectively visualize data is crucial in the contemporary world where information is often voluminous and complex. Visualizations, such as charts, graphs, and maps, provide an intuitive and easily understandable means to interpret, analyze, and communicate patterns, trends, and insights hidden within large datasets. These graphical representations can help researchers, policymakers, and the public to better comprehend and respond to a multitude of issues. In this study, we explore a visualization tool to interpret and understand various data of COVID-19 pandemic. While others have shown COVID-19 visualization methods/tools, our tool provides a mean to analyze COVID-19 data in a more comprehensive way. We have used the public data from NY Times and CDC, and various COVID-19 data (e.g., core places, patterns, foot traffic) from Safegraph. Figure 1 shows the basic view of our visualization view. In addition to providing visualizations of these data, our visualization also considered the Surprising Map. The Surprising Map is a type of choropleth map that can avoid misleading of producing visual prominence to known base rates or to artifacts of sample size and normalization in visualizing the density of events in spatial data. It is based on Bayesian surprise-it creates a space of equi-plausible models and uses Bayesian updating to re-estimate their plausibility based on individual events.