Visualizing conflict in the DRC
The Democratic Republic of Congo is slowly recovering from a conflict known as Africa's first world war, which led to the loss of some five million lives between 1994 and 2003, but many eastern areas are still plagued by violence as various rebel groups continue to operate there
Introduction to the problem
Our team won the 2017 Qlik Qonnections Hackathon that was in partnership with the UN. The goal was to build a conflict analysis tool for the UN to help them better understand conflicts in the Democratic Republic of Congo and potentially even prevent a crisis with early intervention.
We were given a brief that explained the problem and also a dataset of conflict events that happened in Africa over the last 20 years. The dataset revealed the nature of the events taking place and the impact they were having on the lives of innocent civilians.
You can view a demo of the application here
Our team consisted of UX designers, web developers and a data scientist. I was part of UX design team and was responsible for the research and design phases of the project.
Our research phase involved studying the brief, exploring the dataset and learning more about the history of conflicts in the DRC.
We listed specific questions we needed to answer
Who are the main actors and what is their impact on the conflict?
Where and when did these actors come into play?
How and when have civilians been affected by these incidents, in different parts of the country and over time?
What are the relationships between actors and how have they changed over time?
Based on the features and actions of an actor/group, can we predict the likelihood of future behavior?
One of the design challenges here was to not just present facts in the most efficient way possible but to tell a story of what was going on and give people the freedom to explore the dataset and draw their own conclusions. Each of the views focuses on a different type of analysis
Timeline - Storytelling View
Actor Analysis - Exploratory Analysis
Outcome Prediction - Predictive/Actionable View
Timeline - Evolution of Conflicts
The design of this view was driven by wanting viewers to feel the cumulative impact of all of the conflict in the DRC, while giving a user the opportunity to interact with the data and explore it at will.
As time progress you can see a constant shift in the actors as old ones go out and new ones come in. Each of these actors have a devastating impact and have caused a number of fatalities (the white circles), many involving civilians (red circles). By showing this movement of actors including the military force of the DRC, protestors, rioters, various armed groups etc, we wanted to show the reality of the situation is much more complex than just one good vs evil party. When the timeline finally stops at the present time you can see the build up of the past 20 years and the impact of all the actors across the map. You can go back to any point in time to view the events as of that time.
Who are the main actors and where do they have an impact?
In many cases, actors had an impact not just in the DRC but also in surrounding countries. We wanted to use the first half of this view to explore this movement of actors across different countries over time.
What are the relationships between actors?
The second half of this view is about understanding who actors were fighting or allying with and also exploring how these relationship were changing over time. One of the challenges with node diagrams is that they can turn into a hairball that's hard to read. We clustered the nodes by actor type and only displayed actor names on hover to help with this problem.
How likely is a hypothetical event to have a fatal outcome?
To make this application useful we wanted it to be actionable for the UN. The data set we were given contains thousands of conflict events in the DRC, roughly 30% of which resulted in at least one fatality. Given the limited nature of UN resources it is important to understand how to effectively dispatch UN forces.
After creating 100+ new features that describe each event, our data scientist used machine learning algorithms with the most relevant features to predict which events are likely to have a fatal outcome.
The most efficient way to display data isn't always the most effective.
Our applications are consumed by people and sometimes people want to feel a connection and get lost in a story. This was the first data viz project I had done where we incorporated storytelling.
Don't stop at the first few ideas
This project taught me to push myself creatively and think outside the box. It was tempting to stop at the first few boring but perfectly functional ideas especially because we had only 1 week to finish the design.
Collaboration makes everything better
We did a lot of individual and group brainstorming on this project. We built off of each others ideas and that really helped us come up with a design that we were all happy with.