Table 1.

Optimal parameters of different models.

Model	Parameter	Value	Description
BPNN	Number of hidden layer neurons	21
	Activation function	Sigmoid function
SVM	c	0.15	Penalty coefficient
	g	0.75	Parameter of RBF
	Kernel function	Radial basis functions
DT	criterion	Gini	Criterion for feature selection
	max_depth	30	Maximum depth of the tree
	min_samples_leaf	50	Minimum sample number of the leaf node
LSTM	input sequence length	8
	Loss function	Categorical cross-entropy