Kepler.gl is a web-based application for visual exploration of large-scale data sets. Using kepler.gl, a researcher can drag and drop a CSV file into the browser, visualize it with different map layers, explore it by filtering it, and then export the final visualization as a static map or even an animated map. Kepler.gl allows the researchers to map location data, using a point layer, to plot locations of events and places, and, using an arc layer or a line layer, to plot origin-destination correlations. Although beyond my usage in the current exercise, kepler.gl also offers the researcher a way to map aggregate statistics of geographic regions using a grid or polygon layer. Kepler.gl even allows the researcher to plot locations and events using a time playback, to show time and space correlations.
My introduction to kepler.gl involved various visualizations of metadata associated with a dataset of WPA slave narratives created in Alabama during the 1930s. The dataset comprised some 129 interviews of former slaves. While the slave narratives are more famous for their oral and “lived” history of American slavery, kepler.gl enables the researcher to study them geospatially by mapping the location of the interviews and manipulating the data to reveal patterns and correlations in both locations of the interview subjects and of where the interview subjects had been enslaved. I created a point map, a density map, a heat map, a time analysis map, and a line map of the Alabama slave narratives’ interview data. The point map showed the most detail about each interview subject-not only his or her location at the time of the interview, but also determinable biographical data like age, name, and place of birth. The density and heat maps showed clusters of where the interview subjects were located. These maps suggested new lines of inquiry about where former slaves tended to live or at least were in Alabama at the time of the interviews (generally in and around cities). The time analysis map showed the frequency of the interviews’ occurrence according to a timeline, which again raises other possible research questions about whether the interviews took place to accommodate transient interviewers’ schedules. The arc and line graph of the interviews were the most revealing, for they showed that most of the interview subjects had not been enslaved within Alabama or even in adjacent states. The line map thus raises the question of why so many people who had been enslaved elsewhere, particularly farther north, resided in Alabama in the 1930s. In general, kepler.gl functions to prod the researcher to ask quite different questions about a dataset, through paying attention to its geographical aspects as much as more than to its textual content.