A Machine Learning Model to Predict Missing People Status

Abstract :

The phenomenon of the disappearance of people is a global problem, in which factors of security, management of public resources and the emotional aspect related to the loss of a loved one are involved. This paper introduces a model to predict and classify the status of missing persons, using 20 variables referring to the personal and geographical information of the event. The data used was taken from the annual report of the disappeared people published by the technical investigation body of the national prosecutor's office of Colombia in 2017, with 6202 cases. We reviewed scientific literature associated with machine learning models used for the modeling of social phenomena, identifying the most used techniques in these studies. Secondly, the database was debugged in order to proceed with a relational analysis of the variables. Third, three models of supervised data learning are implemented, Decision Trees, Nearest K-neighbors and Random Forest. The results show how that the Random Forest model produces consistently better performance than the other models, over the cross-validation and testing stages