Table 4 The characteristics and ideal meta-parameters of all models developed in forecasting CCRSFm.

From: Developing a novel hybrid model based on GRU deep neural network and Whale optimization algorithm for precise forecasting of river’s streamflow

Main Deterministic Factors

GRU Model (1)

Bi-GRU Model (2)

2GRU× Model (3)

Hybrid 2GRU×–WOA Model (4)

NHN

60

50

60

70

SAF

tanh

tanh

tanh-softsign

tanh-softsign

P-rate

0.4

0.6

0.6

0.5

Optimization Algorithm

Adam

Adam

Adam

WOA

Learning Rate

4E-8

4E-8

3E-8

2E-8

Mini Batch Size

20

20

20

20

Initial Batch Size

128

128

64

32

Convergence Time (s)

48

24

28

33

TLP

11,160

4101

5791

7806