A Logistic Regression and Markov Chain Model for the Prediction of Nation-state Violent Conflicts and Transitions
Date of Award
Master of Science in Operations Research
Department of Operational Sciences
Darryl K. Ahner, PhD.
Using open source data, this research formulates and constructs a suite of statistical models that predict future transitions into and out of violent conflict and forecasts the regional and global incidences of violent conflict over a ten-year time horizon. A total of thirty predictor variables are tested and evaluated for inclusion in twelve conditional logistic regression models, which calculate the probability that a nation will transition from its current conflict state, either In Conflict or Not in Conflict, to a new state in the following year. These probabilities are then used to construct a series of nation-specific Markov chain models that forecast violent conflict, as well as yield insights into regional conflict trends out to year 2024 and beyond. The logistic regression models proposed in this study achieve training dataset accuracies of 88.76%, and validation dataset accuracies of 84.67%. Additionally, the Markov models achieve three year forecast accuracies of 85.16% during model validation. This study predicts that global violent conflict rates remain constant through year 2024, but are projected to increase beyond that timeframe with 95 of the 182 considered nations projected to be in a state of violent conflict from the current 84 nations in conflict.
DTIC Accession Number
Shallcross, Nicholas, "A Logistic Regression and Markov Chain Model for the Prediction of Nation-state Violent Conflicts and Transitions" (2016). Theses and Dissertations. 379.
Applied Mathematics Commons, Other Operations Research, Systems Engineering and Industrial Engineering Commons, Peace and Conflict Studies Commons