UrbViz: Visual Analysis of Urban Indices and Geographically Aware Causality

Abstract

Municipalities and government entities collect massive amounts of daily data from urban activities and events for multiple purposes. Part of this information is used to base their decisions on geographically-targeted budgeting and resource allocation. This requires a deep understanding of the data and how urban indicators impact geographical areas. However, the tools and mechanisms typically used for this decision-making are generally unsuitable, as they focus on small parts of the data or aggregates. This thesis proposes UrbViz, a visual analytics tool that supports interactive geographical analysis of multi-attribute rankings and causality graphs. The system allows to 1) compare and rank different urban areas regarding index scores and urban attributes, 2) identify hotspots and coldspots on the map for a group of selected indicators, 3) find relationships in geographical areas considering scores of indexes and attributes, 4) analyze and explore causal relations between indicators. The effectiveness of UrbViz was evaluated with real-world urban datasets, including traffic collisions, fire incidents, and neighborhood calls datasets, by two domain experts from a local municipality. The results of the study support the advantages of our tool.