The paper presents a visual analytics procedure for analyzing movement data to determine significant places based on recurring events. The procedure addresses problems where relevant places have arbitrary shapes and sizes and must be delineated by processing movement data rather than selected from predefined areas. The workflow consists of four main steps applied iteratively:
Step 1: Event extraction - Relevant movement events (m-events) are identified from trajectories using dynamic attributes representing movement characteristics (speed, direction, acceleration) and relations to spatio-temporal context. Interactive visual query tools enable specification of event-defining conditions. Step 2: Place determination through clustering - Density-based clustering identifies places where events occur repeatedly. Two-stage clustering may be applied: first, spatio-temporal clustering filters occasional events by grouping events concentrated in both space and time; second, spatial clustering unites spatio-temporal clusters sharing spatial positions. A custom distance function accounts for spatial distance, temporal distance, and thematic attributes (particularly movement direction). Interactive parameter adjustment and visual assessment guide refinement until spatially coherent, interpretable clusters emerge. Spatial buffers around clusters define relevant places. Step 3: Spatio-temporal aggregation - Events and trajectories are aggregated by places and time intervals, producing time series of counts and statistics (visit counts, visitor counts, durations, speeds, directions). For trajectory aggregation, flows between place pairs are also computed. Step 4: Analysis - Aggregated data are explored through coordinated interactive visualizations to identify temporal patterns of event occurrences and movements between places. The procedure was demonstrated on traffic congestion analysis (Milan car trajectories) and air traffic analysis (French flight data), revealing temporal patterns and spatial organization of movements.