As an example of geographical regionalization, identifying migration regions is one of the growing areas of study among geographers. Ideally, a system of migration regions can be defined as follows:
"Groups of administrative units with the maximum internal interaction and a minimum flow across the migration-defining boundaries (Ng, R.C.Y., 1969)."
The structures and processes responsible for the formation of regions are not randomly dispersed across the Earth's surface (Duckham et al. 2003). Distance plays a significant role in migration structures, especially on the national scale. However, other factors such as kinship or friendship ties and historical events like Cold War and the Gold rush in the United States also have a significant impact on forming the migration structure in the US over the settlement period. During the nineteenth century, the United States experienced a significant degree of population mobility, as individuals often traveled extensive distances due to the westward progression of European settlement.
To better understand the structure of migration and its evolution over time, we can map migration regions on a large spatiotemporal scale. Studying migration regions through time can tell us how people settled in the United States, what was the structure of flows, and what they can tell us about current social and cultural structures.
The aim of this project is to obtain and visualize the migration regions using migration networks from the large multi-generational Family tree dataset. For this purpose, I plan to use two network community detection algorithm, Louvain and Leiden. By extracting migration regions over time, we can understand the structure of migrations and mobility of people between 1850 and 1920, which occurred between geographically distant places because of their strong migration ties among each other. Moreover, Mapping long-term changes in interstate historical migration flows in the U.S. can tell us about the present cultural and social structures in the country.
I aimed to create a straightforward visualization tool to cater to anyone involved in regionalization studies, enabling effortless map imports and offering comparative visualizations with both visual and statistical analysis options.
For mapping migration regions, I use extracted migrations networks from the family tree dataset, which has been developed by Koylu et al. (2021). The family tree is a network of family members stretching over many generations, containing approximately 40 million individuals. Migration networks are those networks in which the nodes are the states of the US, and edges are the volume of families who moved between states. Using migration networks, including origin, destination, and the volume of families who migrate between states in different intervals and applying the louvain and Leiden algorithm, the regions (communities) for each network can be obtained.
By participating in the usability survey, you will help me understand the strengths and limitations of regionalization mapping and representation techniques. Refer to the 'How to use the maps' section for guidance on utilizing the maps. For further information about the project, its objectives, and the data, please explore the preceding sections.
"Similarity" refers to pairwise similarity, a simple metric used for comparing network structures. This metric works by counting all node pairs found in the same region on both maps, then dividing that number by the total count of pairs.
A higher similarity value indicates that the two maps are more alike in terms of migration regions, meaning more states are situated in the same region on both maps.
Similarity can only be computed for networks (or maps) that have an identical number of nodes (or states).
Modularity is a metric that evaluates the efficiency of a method in forming partitions within a specific network. A higher modularity score suggests that the states within one region have more interconnections in terms of migration.
It's important to note that modularity values are comparable when using the same method.