A Hybrid LSTM-Based Ensemble Learning Approach for China Coastal Bulk Coal Freight Index Prediction

Table 4

The best configurations of the proposed model and benchmark models.

Algorithm	Best configurations	Forecasting cases
		Daily	Weekly	Monthly
LSTM-GBRT	LSTM layer
	Time-lag	8	10	6
	Number of hidden layers	2	2	2
	Number of units in the hidden layers	64	16	32
	GBRT layer
	Number of trees	101	31	41
	Depth of trees	1	1	1
LSTM-RF	LSTM layer
	Time-lag	8	10	6
	Number of hidden layers	2	2	2
	Number of units in the hidden layers	64	16	32
	RF layer
	Number of trees	41	11	21
	Depth of trees	11	11	1
GBRT	Number of trees	191	11	11
	Depth of trees	21	1	1
RF	Number of trees	11	31	11
	Depth of trees	1	1	1
LSTM	Time-lag	8	10	8
	Number of hidden layers	2	2	2
	Number of units in the hidden layers	64	32	64