Table 2. Hyperparameter settings of the forecasting model.

Method	Parameters	Values
LSTM	Time step	5, 10, 15
	Number of hidden layers	3
	Number of units	128, 64, 16
	Activation function	ReLU
	Optimizer	Adam
	Loss function	Mean square error
	Regularization	Early stopping
	Max number of epochs	200
SVR	Time step	5, 10, 15
	Kernel function	Linear
	Cost	10
	Epsilon	0.001
	Cross-validation	5-fold