A Multivariate and Multistage Medium- and Long-Term Streamflow Prediction Based on an Ensemble of Signal Decomposition Techniques with a Deep Learning Network

Table 2

Comparison of models using statistical metrics during the training period.

Model	MAE (m^3/s)	RMSE (m^3/s)	NSCE	MAPE (%)
RBF	39.520	57.504	0.841	28.533
SVR	30.748	47.482	0.892	20.485
RFR	28.692	43.588	0.909	19.891
GRU	25.464	37.813	0.931	18.164
LSTM	23.252	35.091	0.941	16.130
ICEEMDAN-LSTM	15.502	22.162	0.976	12.120
VMD-LSTM	11.579	17.478	0.985	9.514
IVL (ICEEMDAN-VMD-LSTM)	7.083	12.553	0.992	7.465