Skip to main content
. 2022 Aug 16;22(16):6121. doi: 10.3390/s22166121

Table 2.

Evaluation of LSTM hyperparameters.

Optimizer Hidden
Units
RMSE MAPE MAE R2 Time (s)
SGDM 10 5.40 ×102 1.67 ×101 3.85 ×102 0.5226 81.53
20 3.73 ×102 1.12 ×101 2.60 ×102 0.2737 78.02
30 2.32 ×102 7.09 ×102 1.63 ×102 0.7180 71.09
40 2.77 ×102 8.48 ×102 1.95 ×102 0.5981 70.62
50 1.78 ×102 4.82 ×102 1.15 ×102 0.8339 74.71
ADAM 10 5.61 ×102 1.79 ×101 4.08 ×102 0.6419 71.29
20 5.64 ×102 1.76 ×101 4.05 ×102 0.6601 72.94
30 3.33 ×102 1.02 ×101 2.34 ×102 0.4206 74.13
40 2.97 ×102 8.85 ×102 2.06 ×102 0.5396 74.16
50 1.82 ×102 4.84 ×102 1.16 ×102 0.8274 74.43
RMSprop 10 5.13 ×102 1.68 ×101 3.79 ×102 0.3732 72.47
20 5.13 ×102 1.75 ×101 3.90 ×102 0.3762 73.32
30 4.50 ×102 1.51 ×101 3.38 ×102 0.0566 73.74
40 3.10 ×102 9.16 ×102 2.14 ×102 0.4995 75.56
50 3.73 ×102 1.33 ×101 2.92 ×102 0.2722 73.08