Figure 1
2012 ACLED battle incidents
Figure 2
Frequencies of target classes before and after the removal of missing variables
Figure 3
Performance of baseline, naïve Bayes and random forest machine learning algorithms
Figure 4
Performance of learners and random forest classification models
Figure 5
Variable importance for both learners measured in the decrease in accuracy (left) and decrease in overall node purity when variable is removed.
Figure 6
Map of predicted and actual classes for 2012