Density Based k-Nearest Neighbors Clustering Algorithm for Trajectory Data

With widespread availability of low cost GPS, cellular phones, satellite imagery, robotics, Web traffic monitoring devices, it is becoming possible to record and store data about the movement of people and objects at a large amount. While these data hide important knowledge for the enhancement of location and mobility oriented infrastructures and services, by themselves, they demand the necessary semantic embedding which would make fully automatic algorithmic analysis possible. Clustering algorithm is an important task in data mining. Clustering algorithms for these moving objects provide new and helpful information, such as Jam detection and significant Location identification. In this paper we present augmentation of relative density-based clustering algorithm for movement data or trajectory data. It provides a k-nearest neighbors clustering algorithm based on relative density, which efficiently resolves the problem of being very sensitive to the user-defined parameters in DBSCAN. In this paper we consider two real datasets of moving vehicles in Milan (Italy) and Athens (Greece) and extensive experiments were conducted.