Table A.6.

Search space for hyperparameters in LGBM and the best configuration obtained according to the method applied for predicting the TTs of each ambulance.

Method	LGBM search space	Best configuration
		CHRU	CHIH	HNFC
BSTT	max_depth: [1–12]	2	1	1
	n_estimators: [50–1000]	978	820	678
	num_leaves: [31–100]	98	59	99
	learning_rate: [0.001–1]	0.0034	0.0099	0.0157
	subsample: [0.5–1]	0.78	0.68	0.55
	colsample_bytree: [0.5–1]	0.86	0.89	0.92
Proposed methodology	max_depth: [1–12]	6	2	9
	n_estimators: [50–1000]	150	129	401
	num_leaves: [31–100]	94	46	75
	learning_rate: [0.001–1]	0.1888	0.0956	0.0299
	subsample: [0.5–1]	0.72	0.65	0.79
	colsample_bytree: [0.5–1]	0.73	0.76	0.5