The study aims to develop a model for predicting the probability of armed conflict using various data sources.
Utilized an ensemble machine learning approach
Incorporated data from ACLED, UCDP, World Bank, SIPRI, and V-Dem
Employed XGBoost, Random Forest, and LSTM models for predictions
Achieved 87.3% accuracy in predicting armed conflict probability
Demonstrated the effectiveness of ensemble methods in conflict prediction
Abstract
Machine learning ensemble approach to predicting armed conflict using ACLED, UCDP, World Bank, SIPRI, and V-Dem data. Achieves 87.3% accuracy with XGBoost, Random Forest, and LSTM models.