Table 2 Performance of the various models, differing by their machine learning model as well as the provided features. The best performing model is in bold.

Model #	Features				Model Type	ROC AUC	Accuracy	F1 Score	Precision	Recall
	Vegetation, meteorologic, and anthropogenic factors	Fire history	Fire weather indices	Spatio-temporal data
Model 1	v				Logistic Regression	0.798	0.798	0.806	0.801	0.812
	v				XGBoost	0.877	0.877	0.881	0.880	0.882
	v				Random Forest	0.829	0.828	0.828	0.856	0.802
	v				AutoGluon	0.875	0.876	0.880	0.878	0.882
Model 2	v	v			Logistic Regression	0.798	0.798	0.806	0.801	0.811
	v	v			XGBoost	0.887	0.887	0.891	0.890	0.892
	v	v			Random Forest	0.832	0.831	0.831	0.860	0.804
	v	v			AutoGluon	0.885	0.886	0.890	0.888	0.892
Model 3	v		v		Logistic Regression	0.805	0.805	0.812	0.811	0.812
	v		v		XGBoost	0.885	0.885	0.889	0.888	0.890
	v		v		Random Forest	0.831	0.830	0.831	0.856	0.808
	v		v		AutoGluon	0.883	0.883	0.887	0.886	0.888
Model 4	v	v	v		Logistic Regression	0.805	0.805	0.811	0.812	0.811
	v	v	v		XGBoost	0.893	0.893	0.896	0.896	0.896
	v	v	v		Random Forest	0.835	0.834	0.835	0.860	0.811
	v	v	v		AutoGluon	0.891	0.891	0.895	0.896	0.894
Model 5	v	v	v	v	Logistic Regression	0.821	0.821	0.827	0.827	0.827
	v	v	v	v	XGBoost	0.916	0.916	0.919	0.916	0.922
	v	v	v	v	Random Forest	0.846	0.846	0.848	0.867	0.829
	v	v	v	v	AutoGluon	0.913	0.913	0.916	0.914	0.919

Quick links

Search