Fig. 3
From: Predictiveness and drivers of highly pathogenic avian influenza outbreaks in Europe

Summary SHAP plot depicting the global as well as the local feature impact on model 2. Temperature, normalized deviation vegetation index, and normalized deviation water index emerge as the top ranking features. Bio6 and wild positively pushes the prediction of the outbreaks with Cygnus olor being the leading species. Low bio6 (temperature seasonality), mean_temp_q2 values has positive impact on the model (+ ve predictions). On the other hand, reduction in water and vegetation in Q4 and Q1 respectively together with low values in min_temp_q3 has negative impact on the model (− ve prediction).