Visual Analytics-enabled Bayesian Network Approach to Reasoning about Public Camera Data

Abstract

The Visual Analytics (VA) approach has become an important tool for gaining insights on various data sets. Thus, significant research has been conducted to integrate statistical methods in the interactive environment of VA where data visualization provides support to analysts in understanding and exploring the data. However, much of the data explored with VA is inherently uncertain due to limits of our knowledge about a phenomenon, randomness and indeterminism, and vagueness. The Bayesian Network (BN) is a graphical model that provides techniques for reasoning under conditions of uncertainty in a consistent and mathematically rigorous manner. While several software tools for visualizing and editing BNs exist, they have an evident shortcoming when spatial data. In this study, we propose a Visual Analytics-enabled BN approach for reasoning under uncertainty. We describe the implementation procedure using an example of heterogeneous data that includes locations of security surveillance cameras installed in public places.