Uncovering Spatiotemporal Dynamics From Non-Trajectory Data

Michael Oppermann and Tamara Munzner


Design Problem | Paper | Talk | Figures |

Design Problem

In a previous project [Bike Sharing Atlas by Oppermann et al.], we analyzed and compared hundreds of bike sharing networks worldwide, predominantly based on station fill levels that we recorded continuously over a period of 17 months. We illustrated how we can support users with a wide range of expertise to understand and intelligently leverage this type of data in their decision-making. Our interactive visualization can reveal interesting insights, not only into patterns of bicycle us- age but also into underlying spatiotemporal dynamics of a city.

We have begun a new project that is focused on creating visual and predictive decision-support tools centered around building occupancy data. Previously this data has been used for the automatic control of heating, ventilation and air conditioning (HVAC) systems and now, by following the design study methodology, we are opening it up to a broader set of stakeholders in facility management. Initial experiments indicated that making this data accessible and visually explorable can lead to a better understanding how space is actually being used and will ultimately improve space utilization and resource management.

The intriguing underlying similarity between these projects lies in the data characteristics. We are using status changes at distinct locations (non-trajectory), such as the number of available bikes at a docking station or the number of people occupying a certain room, to investigate spatiotemporal patterns. In contrast, much of the previous work has been focused on individual movements (trajectory) or on origin-destination (OD) data.
By noting the similarity in the data, we can take what we learned in both projects to discuss general implications of spatiotemporal non-trajectory data in terms of ethics, data preprocessing, tasks, and visual encodings. Our goal is to generalize our findings in the context of urban data visualization with the hope to inspire other researchers and designers.

Short Paper

Uncovering Spatiotemporal Dynamics From Non-Trajectory Data
In Proc. 2018 CityVis Workshop, pp. 4-6, 2018.

Pre-Print PDF
CityVis Workshop

Talk

Michael Oppermann presented this work on Oct 22, 2018 during the CityVis workshop at IEEE Vis 2018.

High-Resolution Figure

Fig. 1. The visualization of average docking station fill levels exposes bike-sharing commuting behaviors in different cities around the world.




Last modified: Apr 28, 2019.