Exploratory vs. Model-based Mobility Analysis

Authors

• Jussi Nikander
• Tanja Kantola Department of Real Estate, Planning, and Geoinformatics, School of Engineering, Aalto University, O.O. Box 11200, FI-00076 Aalto
• Kirsi Virrantaus Department of Real Estate, Planning, and Geoinformatics, School of Engineering, Aalto University, O.O. Box 11200, FI-00076 Aalto

Keywords:

explorative spatial data analysis, visual analytics, clustering, multi-variate visualization, knowledge, spatial analysis process, crisis management

Abstract

In this paper we describe and analyze a visual analytic process based on interactive visualization methods, clustering, and various forms of user knowledge. We compare this analysis approach to an existing map overlay type model, which has been developed through a traditional modeling approach. In the traditional model the layers represent input data sets and each layer is weighted according to their importance for the result. The aim in map overlay is to identify the best fit areas for the purpose in question. The more generic view is that map overlay reveals the similarity of the areas. Thus an interactive process, which uses clustering, seems to be an alternative method that could be used when the analysis needs to be made rapidly and utilizing whatever data is available. Our method uses visual analytic approach and data mining, and utilizes the user knowledge whenever a decision must be made. The tests carried out show that our method gives acceptable results for the cross-country mobility problem, and fulfills the given requirements about the computational efficiency. The method fits especially to the situations in which available data is incomplete and of low quality and must be completed by the user knowledge. The transparency of the process makes the method suitable also in situations when results based on various user opinions and values must be made. The case in our research is from the crisis management application area in which the above mentioned conditions often take place.