Discovering Local and Global Co-Location Patterns in Trajectories with Different Properties

Most trajectory mining approaches consider a very small set of properties to extract patterns from trajectories. Besides, a lot of them consider those properties separately. In this paper we present a method to find co-location patterns based on different properties of trajectories along time. The proposal allows to identify a sequence of co-locations composed by different properties (distance, acceleration, speed, time etc) that represents the behavior of a set of trajectories. The sequences of co-locations patterns present the evolution of an event, for example traffic jam. By using this knowledge is possible to anticipate the phenomenon occurrence, and to take actions to solve problems regarding to the event.