The preliminary results showed that random forest worked best for the PB imbalanced data, having a 0.84 weighted average in precision and recall metric. The model reproduces 9 of the PB with low error 4.5% and overestimates ...