Kepler.gl lets researchers, as the website says, “render large datasets quickly and efficiently.” Indeed, this open-source software lets the user drag and drop large datasets, apply filters, and apply layers to view the data through points (to mark specific locations) or arcs and lines to track movement. The tool lets the user create geospatial visualizations without having to code.
At Kepler.gl, a “Get Started” link takes new users to a demo version where the researcher can try mapping different datasets. After opening the demo, a screen where the user can drag and drop a data set appears. Creating a dataset is probably the hardest part unless you have one that is ready to drop in. Kepler.gl works with CSV, GeoJSON, and kepler.gl Json files. The table must include columns labeled latitude and longitude containing spatial data. According to the site, “Geometry coordinates should be presented with a geographic coordinate reference system, using the WGS84 datum, and with longitude and latitude units of decimal degrees.”
Once you have uploaded your data to Kepler.gl, you can create a point map that shows the occurrence of an event at a given location. Under layers, choose “point.” As you roll your cursor over the points, additional data from your table appears for each site. With radius, the user can choose to make the size of their points larger or smaller.
Multiple points at a single location appear as one point. To show the density and proximity of events to each other, the user can change to a heat map or cluster using the basic map selections.
Kepler.gl also has features to view a category map by sorting according to the data table’s specific categories. I ran a trial using data on the ex-slave interviews conducted in Alabama from the Library of Congress WPA Slave Narrative Collection. The researcher can select any category in the data set to see the prevalence of different groups at different locations. In the slave interviews, for example, the visualization shows that house slaves made up a much higher proportion of the interviewees than they would have made up in the general population of enslaved people. One quickly realizes how data visualization can make outliers and patterns knowable.
There is also a time scale map that shows how close together in time different events occurred. The researcher will need a date category in the data table to do this. For the formerly enslaved, the viewer sees that most Alabama interviews were done within a few months of each other and in the summer months. Kepler.gl’s bar graph displays the number of points at each time and calculates the span automatically.
Finally, a network map can be created to show related points on a map to be linked by arcs or lines. This data includes both the location where an interview took place and in most cases locations at which an interview subject was enslaved. The map below shows the networked Kepler.gl map of the Alabama slave narratives. The purple dots indicate the locations where people were enslaved. The green line indicates their movement away from the location of enslavement, and the yellow line brings them to the place where they were interviewed. The user can select lines instead of arcs. Compared to the point map, these visualizations provide a good idea of the amount of movement freed people insisted on after the Civil War, even if they do not show intermediary stopping points along the way.