Flow Mapping with Graph Partitioning and Regionalization
Flow Mapping with Graph Partitioning and Regionalization is an integrated software tool to explore flow patterns in large spatial interaction data. It involves two packages: (1) GraphRECAP, which uses spatially constrained graph partitioning to find a hierarchy of natural regions defined by spatial interactions; and (2) FlowMap, which visualize flows based on the discovered regions and related attributes. In both steps, the original flow volume is transformed to a modularity measure, which is the difference between the actual flow and the expected flow. Expected flows can be calculated based on the original flow matrix or the population in each region / place. The tool allows filtering flows by setting a threshold or exploring flows at different region levels. Multivariate information for each flow may also be used for multivariate mapping.
Following are example maps created with the migration data. The geographic space can be partitioned into communities (areas) based on the flow network and thus we can map flows at different hierarchical levels.
(Made with the new version of FlowMap, which is to be released soon.)
- Guo, D. (2009). "Flow Mapping and Multivariate Visualization of Large Spatial Interaction Data", IEEE Transactions on Visualization and Computer Graphics, 15(6), pp. 1041-1048 [Journal Link]
- Guo, D. (2009). "Greedy Optimization for Contiguity-Constrained Hierarchical Clustering", The Fourth International Workshop on Spatial and Spatiotemporal Data Mining, IEEE International Conference on Data Mining (ICDM 2009), Miami, FL.
Java is needed to run the following software. You can verify if Java is already installed on your computer at this link: http://java.com/en/download/installed.jsp
We will release a new version for both GraphRECAP and FlowMap. Please check back soon for much improved software tools.