A visual summary of spatio-temporal events that preserve áreas and neighborhoods in a 2D plot
Carregando...
Data
2021-12
Autores
Orientador(res)
Poco, Jorge
Métricas
Título da Revista
ISSN da Revista
Título de Volume
Resumo
The analysis of time-evolving clusters is an essential task in the study of spatio-temporal
data. People often use these clusters to represent events automatically detected in many
fields, such as human mobility and disease outbreaks. Performing visual analysis of this
data type is challenging for current state-of-the-art techniques due to factors such as spatial span covered by clusters, spatiotemporal intersections, and temporal evolution. All of this makes widely used geographical map-based techniques suffer from overplotting and cluttering, therefore, ineffective for this purpose. Visualization techniques used to analyze results use animation or interactivity to represent the three dimensions, but they show limitations on interpretation. To overcome these limitations, we present Events-Vis,
a method for visualizing spatio-temporal clusters event data in a static temporal plot by representing the space in one dimension. We linearize the space using two strategies: a greedy algorithm and a convex optimization. In both cases, our goal is to preserve neighborhoods and intersections. We demonstrate the effectiveness of our method in a
series of experiments and a case study using both synthetic and real-world datasets.
Descrição
Trabalho de conclusão de curso