Gravity Model

After looking into Dr. D’s guide to gravity models, and the Garcia et al. paper, I learned a great deal about gravity models and their applications. I have done a lot of research on gravity models in past years, specifially surrounding disease transmission, and know that the models work well to determine the interaction between distinct locations, as well as the movement of individuals from different locations. Dr. D’s guide focused on London, and we were able to create a flow map, which showed the how individuals living in London moved from home to their place of work, as well as their main method of transportation. The image below shows the formula used for the gravity model, where Pi is the population of location i, Pj is the population of location j, and Dij is the distance between location i and location j. In his paper, Garcia introduced GTSIM, which is a gravity model that also has layers including socioeconomic, demographic, and envrionmental characteritics. Many of these extra layers are obtained using census and population data. These layers can be very helpful in getting a more comprehnsive understanding of how and why individuals are moving from one location to another. Garcia also touches on push-pull factors, explaining that specific locations often have either push factors, which encourage individuals to leave the locations (high crime rates, poverty), or pull factors that encourage individuals to move to specific locations (high employment rates, closer to transportation networks). These factors have a very large influence on where individuals choose to live and work. Garcia’s paper concluded the study by acknowleding that although the GTSIM models were able to capture migration patterns of the different countries realitvely well, they were not able to capture them holistically. In this case, it may be more beenficial to use more than one method instead of just gravity models to better understand migration patterns.

The image above is a plot that represents some of the urbanized, highly populated areas in Lira (similar to project 2). This plot is different from the plot produced in project 2, however, because we have layered the synthetic population (created in project 3) ontop of the urbanized areas.

Using the centerpoints along with internal migration flow data, I was able to create a plot of migration patterns for a specific subdivision in Uganda, called Lira. Using voronoi polygons, we were able to identify the centerpoints of some of the primary, more populated areas located around Lira, Uganda. The plot below represented points that gave us a relatively accurate description of the origin and destination points (and the path between them) in the subdivison Lira. Expanding our view to the country as a whole, as you can see from the plot above, there were approximatley 58 destination points around all of Uganda. The internal migration flow data that I received from WorldPop was from 2010. Since this data is relatively outdated, I might have gotten a more accurate representation of the movement around Uganda by using more recent migration data. In the future, it may be beneficial to include more comprehensive data, including information regarding where work buildings, hospitals, train stations, airports, and schools are located. By identifying these variables, we may be able to better understand where individuals are moving day to day.

OD Matrix

Above is an image of my Origin-Destination Matrix. The top row (x axis) represents origins, while the columns (y axis) represent destinations. The boxes with NA represent the migration from an origin point that is the SAME as the destination point (that is why they do not need to be included). The boxes without NA (numeric vaules) represent the predictive value of migration from one origin (row) to a destination (column). Lastly, below I have included an animation of movement around the country of Uganda. My animation only shows the movement of one point from the origin point to a destination. In the future, I hope to be able to create the same animation with multiple points moving across Uganda. This animation was fun to make! Gravity models are a very advanced data science method and it was cool to be able to make one for the country I chose! I believe that gravity models can be extremely benefical to track the movement of individuals for a number of different reasons. Past research has found that gravity models are helpful in determing the effects of natural disasters (where individuals move after a large natural disaster), the spread of diseases (where individuals with a disease move, where they spread the disease), and the interaction between different areas (trading). These are just a few reasons (among many!) of occrurences where gravity models can be very helpful.

The Animation!