Introduction
Perhaps digging into this subject matter might be analogous to “doom scrolling”, but with the frequency at which news syndicates push crisis, death, and conflict, and then quickly abandon them for another inevitable breaking story, I miss the context of time. This time can provide historical orientation and allow for the artifact of change to be proportionately represented. In setting the lens wide with a global view, an additional dimension of context can be facilitated through natural comparison of geolocation adjacency.
Thus informed my approach to keep a longer view on the geographic data, set the stage if you will, that would allow the viewer’s curiosity to lead them in exploring the data. My goal was to use Tableau Public, a data visualization software that allows users to publicly share their creations for free, to spatially display world data and give control to the viewers to move through the data.
Inspiration
I took inspiration from two realms: design elements and animation functionality.
When searching for inspiration for the design aspect of the visualization, I was drawn to the color choices and simplicity of the visualization called “Diverse Diners” I found on Tableau Public’s discover page.
Secondly, I was interested to explore methods to display space and time together. This is a notoriously difficult challenge, but I took a “simple is better” approach to communicate the data moving through time.
The GIF on the right (click image to view) demonstrates inspiration for the possible solution of the space/time conundrum. I appreciated its simplicity and wanted to replicate in my final output.
Lastly, this visualization brings the viewer along in a 5 minute journey through 4500 years of “every battle referenced on Wikipedia (10,624)”. As a temporal map, I found the presentation pulled you in as a viewer in a way that a movie might pull you in when you don’t quite know all the details that lead up to a popular event. Sitting and watching dots on a map pop up seems mundane, but it was actually quite engaging (the music probably had something to do with it).
Materials | Data
I had a specific set of criteria when it came to identifying the dataset I was going to use for this project. First, the spatial data points had to be in a standard format. In the case of the dataset I chose, the spatial indicators were in latitude and longitude. Secondly, the origin of the data (for the most part) would be rooted in some neutrality (collected with the intent of spreading awareness). The caveat is that nothing is really neutral, but given the background on the chosen company, the choice was relatively sound.
Dataset
I used data from the non-profit organization with 501(c)(3) status, ACLED, which is a “Disaggregated Data Collection, Analysis & Crisis Mapping Platform. ACLED collects real-time data on the locations, dates, actors, fatalities, and types of all reported political violence and protest events around the world.”
I utilized their data export tool as seen below to export and download the data from their site.
Due to the large amount of data being aggregated, the site recommended pre-filtration prior to downloading of the .csv (comma-separated values) file. I limited the time frame to 2015 to April 8, 2022 (which at the time of download, was the most current date available). I only selected three event types: Battles, Explosions/ remote violence, and Violence against civilians. I kept the filters for region/country/location all open so the data returned would be inclusive of all those values.
The resulting data file (.csv) was about 283 MB in size with 551,432 rows of events described by type of event, time of occurrence, location, actors, fatalities, and notes. This file/dataset was then uploaded to Tableau to be visualized. This was the only dataset used for this project. For the main visualization, I made the decision to highlight latitude and longitude for place (i.e. a map), the type of event, and fatalities. Additional fields would be displayed in the “tool tip”, a box of text that appear when the viewer mouses on a particular data point on the map in Tableau.
Visualization Methods
Colors & Fonts
The color choices arguably took the longest time to decide, as great consideration went into pulling in colors that would illicit certain psychological effects. A warm color palette provided the perfect color scheme to depict heat, anger, trouble, and something “bad” ( as red tends to denote in many visualizations).
Battles = Orange| Inference to fire, edging its way to hotter red
Explosions..= Red | Incendiary reference, analogy to fire
Violence..= Yellow | relation to fire, but visually on the map surrounds the climactic points of Battles and Explosions.
I also decided that the background itself needed to play a role the map narrative. I chose a “dark” map to first provide greater contrast to the warm colors of the data points and second to give a sense of unknown, night time, night vision, maybe even a control panel from a military device.
Lastly, the color for the fatality measure layer was a challenging choice to make. Ideally it needed to provide enough contrast to stand out against the event data so it wouldn’t get lost in the overlap of event dots. White seemed like the best choice as it stood out and also provided the inference of removal or “whited out” (as death most definitely is).
Tool Tip Choices
Since much of the map was graphic in nature, I thought using the “tool tip” filled with additional data as text would compliment and provide more context to the visualization. The chosen font, Courier, reminded me of something an old computer or even a typewriter would produce, which had a very impersonal feeling. The resulting “tool tip” (fig. 5 ) almost looked like something found in government military files.
Tableau Map Controls and Functions
I was interested in trying out the mapping and temporal features in Tableau Public had to offer through “Map Layers” (geographic layers for maps give you the ability to turn on and off dimension measures to augment the user experience and provide more depth of information) and also “Pages” (which enables the creator of the viz to cycle through “slides” of the map populated with a certain dimension and filter applied).
By enabling the Tableau map options such as shown to the right, the viewer has greater ability to zoom in and target geographic areas of their choosing through search, pan, and zoom. The search function is especially useful because the view can directly enter the country in the search to zoom in the map to that view.
Results
Overall, I believe the end result accomplished what goals I had set out to complete. The map visualization is relatively simple, focusing on only 2 dimensions. I experimented with the map functionalities in Tableau, which “Map Layers” allowed 2 dimensions of data to exist on the same map with 2 different color scales. The “Slides” feature did enable a “click through time” experience for the viewer, however, the front end user experience was a bit cumbersome. I opted to experiment with other modes to communicate “time” in a more compact manner such as looping .png files together to create a time-lapse video file and also an animated GIF.
Experimentation with Animation
In order to clean up the image and prepare a succession of them to be stitched together to create an animation, I used Adobe Photoshop (an image editing software) to reorganize the legends, adjust the white space around the map and decrease the image size. Since I needed the Maps to stay aligned and consistent from frame to frame, I leveraged the “actions” functionality in Photoshop. By selecting this function prior to me taking all the editing steps I mentioned above, the program recorded the steps I took to perform them (i.e. which drop downs I selected, the size of the canvas adjustment, the resolution adjustment, etc.). Ultimately this saved me a lot of time and allowed for the exact same treatment for each frame. The only slide I did anything different to was the first slide where I moved the legend location and kept them on that slide. For the other slides, I decided not to include the legend as the viewer already got this info and I wanted them to focus on the changing placement of the data.
To make the animated GIF, I used the website GIPHY. The site accepts multiple file types: JPG, PNG, GIF, MP4, or MOV. I ended up making the animated GIF from the MP4 video file that I created in Photoshop.
A data time lapse visualization of the amount of battles, explosions/remote violence, and violence against civilians through 2015 to 2022 globally. Total fatalities are indicated by the size of the circle (larger radius = higher amount of fatalities)
I additionally tried making an animated GIF, but I was disappointed with the resolution. It also felt slightly disruptive to jump to another window while reading the report in order to see the GIF.
Analysis
The hot spots of violence seemed to be contained to similar areas on the map when looking year over year. However, I did notice that some countries didn’t always appear in the data consistently. For example, Central America doesn’t start to “light up” until the year 2018, which from that point in time until the end of the visualization’s date range in 2022, remains strongly populated with events and even deaths.
Reflection
Despite the richness of the dataset, I decided to not explore trough data analysis, but instead explored the newer map functionalities in Tableau and to think more intentionally about the design output. Because of that decision, I tallied up shortcomings I encountered with the platform. First, since there was a lot of data, the map load times in Tableau were lengthy, which I surmised also affected the map and map layer navigation of the user interface via slower response times when the user made navigation selections. In retrospect, the tool’s function as a data visualization (especially if the data is tidy) tool supersedes the design functionality. At times, I had difficulty in making slight adjustments to arrange legends, coloring, and layouts.
To advance this project, I would want to add additional data context through charts and graphs (like the “Diverse Diners” viz mentioned in my inspirations) to tease out interesting patterns and create a deeper narrative with all the available data in the dataset.
However, something that pulls me even stronger is that of the psychological and social aspect of the data: with the hyper focus on “humans being bad”- how does that move the general psyche of human consumers of this data? Is the negative a primary conduit for (the hopes of) change?