Interpretable Predictions of Tree-based Ensembles via Actionable Feature Tweaking

Machine-learned models are often described as "black boxes". In many real-world applications however, models may have to sacrifice predictive power in favour of human-interpretability. When this is th