Table 5.

Evaluation of different variations of LSTM model with sliding window algorithm.

Optimizer	Neuron count and number of LSTM layers	Activation function	Estimated Root Mean Square Error (RMSE)
SGD	128, 128, 128	ReLU, ReLU, sigmoid	0.0722
Adam	128, 64	ReLU, ReLU	0.0693
Adam	128	ReLU	0.0764
Adam	32	ReLU	0.0877