Computer and Robot Vision
Data Mining: Practical Machine Learning Tools and Techniques, Second Edition (Morgan Kaufmann Series in Data Management Systems)
Spatial clustering of structured objects
ILP'05 Proceedings of the 15th international conference on Inductive Logic Programming
| Hi-index | 0.00 |
This paper presents a clustering approach which groups German municipalities according to mobility characteristics. As the number of measurements for nationwide mobility studies is usually restricted, this clustering provides a means to infer mobility information for locations without measurements based on values of their respective cluster representatives. Our approach considers local and global information, i.e. characteristics of municipalities as well as relationships between municipalities. We realize previous findings in urban geography by using techniques from graph theory and computer vision. Our clustering consists of a two-step model, which first extracts and condenses single mobility characteristics and subsequently combines the various features. We apply our model to all German municipalities between 10,000 and 50,000 inhabitants. The clustering has been successfully applied in practice for the inference of traffic frequencies.