Academic coffee: Using AI to find the graves of Mexico’s missing people

This is based on Quartz’s article “Machine learning is being used to uncover the mass graves of Mexico’s missing”. If you want to know more click on the link.

Mexico is a country were at least 30,000 people have disappeared since 2006. Sadly the search for most of these missing persons often has to start underground.

Until recently the large size of Mexico’s territory made it really hard to think where to start looking for someone. A team of multi-country researchers, data scientists, and statisticians is using machine learning to predict which counties in Mexico are most likely to have hidden graves. A team which is composed of three groups: the Programa de Derechos Humanos at the Ibero-American University, Data Cívica and the Human Rights Data Analysis Group (HRDAG).

Each one helps with some specific piece of analysis and together help to form the bigger picture. They have created a database with details of every report of graves found. They have been able to create a profile of sociodemographic data for every one of the 2547 counties in Mexico.

“Patrick Ball, HRDAG’s Director of Research and the statistician behind the code, explained that the Random Forest classifier was able to predict with 100% accuracy which counties that would go on to have mass graves found in them in 2014 by using the model against data from 2013.” It could also predict which counties did not have a hidden graves on them!

Counties with hidden graves are likely to have a lower average income than other counties, be more rural than urban, have higher numbers of indigenous residents. Also, many of the counties have been found to have strong connections to drugs and high homicide rates. The counties with found graves tend to have highways and they also evidence a pattern of being close to borders. The team at Data Cívica reports that three out of every ten disappearances happen in the states of Tamaulipas or Guerrero.

“Though the team has been able to use data from 2013 to predict accurate results for 2014, they haven’t yet been able to do the same for 2017. Before this can be done, Núñez’s team has to update their database with media mentions from 2016, a task that is forthcoming and time-consuming.”