I Predict a Riot!

Predicting the future is here! Well, maybe not quite, but data scientists are currently working hard to accurately predict rebellions, ethnic violence, insurgencies, and mass atrocities around the world through the use of supercomputers and algorithms. With the current political situations and turmoil in countries like Ukraine, Venezuela, Thailand, Syria, and Egypt, making these types of accurate and reliable predictions can be very valuable information.

Image
Protests in Tahrir Square, Egypt.

So far there have been several attempts to calculate these types of predictions.  For instance, Duke University’s Ward Lab collects and deciphers big data using several software programs that analyze news articles from around the world in order to make predictions about which countries are most at risk for rebellion, increase in insurgency, ethnic violence, domestic crisis, and international crises.  Ward Labs have seen some success in accurately making these predictions. In July they estimated that Paraguay had a 97% chance of insurgency, which later proved to be accurate when guerilla attacks increased.  Additionally, in October, Ward Labs predicted that Thailand was at a 95% risk for a domestic crisis, and in December predicted that Iraq’s probability of having an international crisis was 99%; both predictions have since proven to be true. Ward Labs has not been able to predict everything, though; the current crisis in Ukraine was not on their lists until after the protests had begun.  This is okay, however, as they maintain that their main objective is to test theories, and making accurate predictions is a difficult thing to do.

Image
(Source: USAID.gov)

USAID is even getting in on the action, partnering with Humanity United to create the Tech Challenge for Atrocity Prevention.  This worldwide technology competition challenges participants to use current technology to create innovative ways of preventing atrocities.  One of the challenges in the competition is to create a model or algorithm to predict where future atrocities are most likely to occur.  Xiaoshi Liu won this challenge, earning himself $12,000 USD.  His algorithm creates decision trees for every five day period, using data from the most recent atrocity and social-political records.  If his model is indeed effective, it will help USAID effectively calculate what countries need help most and mitigate any damages.

Helping data scientists make these predictions is Kalev Leetaru, creator of the Global Database of Events, Language, and Tone (GDELT).  GDELT is a database that stores information about political events from around the world.  Listing who did what to whom, when, and where, GDELT has recorded more than 200 million events going back to 1979, and plans to go even further back to the year 1800.  Every day, information is gathered by examining news reports from all the countries in the world, and through sentiment analysis – a computer automated method to determine the attitude of the writer or speaker – GDELT is able to catalogue human behavior and beliefs across the world. GDELT is available to the public, and has since provided many political scientists the information with which to test their theories and make predictions about future events.

Image
Mapping areas of protest in Ukraine
(Source: gdeltproject.org)

Making accurate and reliable predictions has proven to be hard; there are always going to be variables that are impossible to envision, such as plane crashes with political leaders on board, or natural disasters that wipe out cities.  We may never be able to make perfect predictions, but with the growing popularity of the business, we are certainly getting better at it.  There is a future for predicting the future.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s