An R Package for using the space-time prism concept on trajectory data

Organisator Laboratory of Geo-information Science and Remote Sensing

wo 27 september 2017 14:30 tot 15:00

Locatie Atlas, building number 104
Droevendaalsesteeg 4
6708 PB Wageningen
Zaal/kamer 2

By Mark ten Vregelaar

Analysis of animal movement is fundamental for understanding the dynamics of populations, their interactions among each other, and their interaction with the environment. The movements of humans and animals can be tracked using a range of location-aware technologies. The resulting trajectories consist of successive control points in space and time. Trajectories can have great scientific value to time geographers, urban researchers, transportation analysts, social scientists, and ecologists. Space-time prisms are a representation of where an individual may have been in between the control points of a trajectory. A space–time prism is the sub-space an individual can reach, given a speed limitation and a pair of control points. The space-time prism concept can be used for a variety of applications. In this thesis research, functionality for handling space-time prisms in R was developed and bundled in the package “STPtrajectories”. This package allows users to visualise space-time prisms in 3D, calculate the Potential Path Area (PPA) of a trajectory, generate random trajectories, and calculate when an individual may have been at a spatial location. The package also contains a method to test if individuals may have met each other. The latter is done with the alibi query, a Boolean query that tests if space-time prisms intersect. Users can also retrieve the potential time span and the Potential Intersection Area (PIA) of the found encounters. To test the usability of the created package the user-friendliness and computation times of the final R package were examined. A trade-off between accuracy of the outputs and the computation time of the methods was found. The effect of the inputs on the computation time follow logical trends and can be explained by the methodology used to generate the outputs. The documentation of the package was judged helpful; it was found to include useful examples but less detailed than similar R packages.