Interactive Analysis of Event Data Using Space-Time Cube Peter Gatalsky, Natalia Andrienko, and Gennady Andrienko Fraunhofer Institute for Autonomous Intelligent Systems Schloss Birlinghoven, D-53754 Sankt Augustin, Germany http://www.ais.fraunhofer.de/~gatalsky {gatalsky@ais.fraunhofer.de, gennady.andrienko@ais.fraunhofer.de} Abstract In exploratory data analysis, the choice of tools depends on the data to be analyzed and the analysis tasks, i.e. the questions to be answered. The same applies to design of new analysis tools. In this paper, we consider a particular type of data: data that describe transient events having spatial and temporal references, such as earthquakes, traffic incidents, or observations of rare plants or animals. We focus on the task of detecting spatio-temporal patterns in event occurrences. We demonstrate the insufficiency of the existing techniques and approaches to event exploration and substantiate the need in a new exploratory tool. The technique of space- time cube, which has been earlier proposed for the visualization of movement in geographical space, possesses the required properties. However, it must be implemented so as to allow particular interactive manipulations: changing the viewing perspective, temporal focusing, and dynamic linking with a map display through simultaneous highlighting of corresponding symbols. We describe our implementation of the space-time cube technique and demonstrate by an example how it can be used for detecting spatio-temporal clusters of events. Keywords: space-time-cube, spatio-temporal data, exploratory data analysis, data visualization 1. Introduction This paper deals with some issues of exploratory analysis of spatio-temporal data, that is, data having both spatial and temporal references. In particular, we focus on the analysis of events – discrete spatial objects having relatively short life duration. Examples of events are earthquakes, traffic incidents, occurrences of diseases, or observations of rare animals. An important question arising in analysis of event data is how the events are distributed in space and time. A proper data representation is needed for answering this question. In early 70s Hägerstrand [1] developed a graphic view on time as an additional spatial dimension. He suggested a three-dimensional diagram, the so-called space-time cube, to show life histories of people and how people interact in space and time. The base of the cube represents the two-dimensional geographical space while the cube’s height represents the temporal dimension. At the time when the concept was introduced, the options for creating such graphics were limited to manual methods. This was an obvious limitation of the approach, since creation of each new diagram was a laborious drawing exercise. Today, when modern computer technologies provide much better opportunities for data visualization, the interest of researchers to Hägerstrand’s time-geography has revived [2]. Prominent application examples have been made in ITC, Enschede, the Netherlands [3]. Thus, Figure 1 shows the representation of the Napoleon’s Russian campaign in 1812. Summarizing all experiments carried out with space-time cube, a project of dynamic visualization environment has been proposed [4]. Figure 1. Representation of object movement in a “space-time cube”. (Source:[4]) Both Hägerstrand and the research team from ITC applied space-time cube to data concerning movement of objects, i.e. changes of spatial locations. In this paper, we Proceedings of the Eighth International Conference on Information Visualisation (IV’04) 1093-9547/04 $ 20.00 IEEE