Modelo para la detección de anomalías sobre objetos en movimiento en bases de datos espacio temporales

Abstract

In this thesis the design and implementation of a model for the path anomaly detection from segmentation and discovery of movement patterns in geographic objects that undergo a displacement in the time, the subject is approached in such a way that the anomalies are points or segments of the trajectories instead of the complete trace taking into account the multidimensionality of the data (space, time, speed), in addition, the approach of the model is made on the support of spatio-temporal databases, especially those of objects in movement that allow the trajectories to be stored in a more friendly and efficient, also taking advantage of the capabilities to perform queries and operations much easier than in traditional relational databases or other formats of records. Finally, the proposed model is verified using real data from the system mass transit of Bogotá - Transmilenio to validate its operation, but the model is much more general, which means that it can be adapted in many more applications in systems where there are moving geographic objects