Chinese Academy of Sciences

(a) Main visualization on anomalous regions (b) Zoomed-in view with anomaly bars Figure 1: An overview of the interactive spatiotemporal anomaly analytic tool. In the main visualization interface, filters (black box) and time slider (white box) allow interactive exploration of the evolution of multi-dimensional attributes. Anomalous activities are visualized through 2D grids (left) and 3D bars (right) through geographic information system. As we move into big data era, many data grows not just in size but complexity with a rich set of attributes that contain location and time information, such as data from mobile devices (e.g., smart phones), natural disasters (e.g., earthquake and hurricane), epi-demic spread, etc. We are motivated to build a visualization tool for exploring generic spatiotemporal data, i.e., records containing time location information and numeric values. Since the values evolve over time and across regions, we are particularly interested in de-tecting and analyzing the anomalous changes over time/space. Our visualization tool is developed based on one popular geographic information system (i.e., Google Earth) and uses a simple yet ef-fective technique, i.e., 3D bars for representing the value dynamics. By combining data mining algorithms such as GridScan as an over-lay 2D grids on top of map, the tool may guide users effectively to find potential anomalies.