A Visual Analytics Toolkit for Cluster-Based Classification of Mobility Data

Abstract

In this paper we propose a demo of a Visual Analytics Toolkit to cope with the complexity of analysing a large dataset of moving objects, in a step wise manner. We allow the user to sample a small subset of objects, that can be handled in main memory, and to perform the analysis on this small group by means of a density based clustering algorithm. The GUI is designed in order to exploit and facilitate the human interaction during this phase of the analysis, to select interesting clusters among the candidates. The selected groups are used to build a classifier that can be used to label other objects from the original dataset. The classifier can then be used to efficiently associate all objects in the database to clusters. The tool has been tested using a large set of GPS tracked cars.