We propose a visual analytics procedure for analyzing movement data, i.e., recorded tracks of moving objects. It is oriented to a class of problems where it is required to determine significant places on the basis of certain types of events occurring repeatedly in movement data. The procedure consists of four major steps: (1) event extraction from trajectories; (2) event clustering and extraction of relevant places; (3) spatio-temporal aggregation of events or trajectories; (4) analysis of the aggregated data. All steps are scalable with respect to the amount of the data under analysis. We demonstrate the use of the procedure by example of two realworld problems requiring analysis at different spatial scales.KEYWORDS: movement, trajectories, spatio-temporal data, spatial events, spatial clustering, spatio-temporal clustering
INTRODUCTIONMovement data (also called mobility data) describing changes of spatial positions of discrete mobile objects are nowadays collected in growing amounts by means of current tracking technologies, such as GPS, RFID, radars, and others. Automatically collected movement data are semantically poor as they basically consist of object identifiers, coordinates in space, and time stamps. Despite that, valuable information about the objects and their movement behavior as well as about the space and time in which they move can be gained from movement data by means of analysis [2]. Movement can be viewed as consisting of continuous paths in space and time [18], also called trajectories, or as a composition of various spatial events [3]. As noted in [6] and [30], there are many definitions of the term event. We adhere to Kim's definition of events as exemplifications of properties or relationships at some times [22]. Spatial events are events localized in space [6].The event-based view of movement is particularly suitable for applications and tasks where analysts are interested in occurrences of certain movement characteristics such as very high or very low speeds. Each occurrence is a spatial event. Events that are relevant to the goals of analysis need to be extracted from movement data. Such events will be further called movement events, or m-events.There is a class of problems where analysts need to determine places in which m-events of a certain type occur repeatedly and then use these places in the further analysis. For example, having tracks of multiple cars in a city, a traffic analyst may first need to find places where traffic jams occur and then investigate in which times of the day they happen and how long they last. From trails of migratory birds, an ornithologist may wish to extract places where the birds stop for resting and feeding and then analyze the temporal patterns of visiting these places and travelling between the places. We point out that relevant places can only be delineated by processing movement data, that is, there is no predefined set of places (e.g., compartments of a territory division) from which the analyst can select places of interest. The relevant p...