Table 4 The results of the number of neurons in hidden layer (4–8 neuron).
Structure of ANN | R (all) | R (test) | R (training) | R (validation) | MSE | MAPE | Optimum epochs | Stop validation | Gradient range |
---|---|---|---|---|---|---|---|---|---|
3:8:1 | 0.902 | 0.95 | 0.889 | 0.918 | 0.0011 | 0.074 | 4 | 10 | 10–2–10–4 (0.0008) |
3:7:1 | 0.936 | 0.934 | 0.929 | 0.975 | 0.0006 | 0.063 | 2 | 8 | 10–3–10–4 (0.0001) |
3:6:1 | 0.932 | 0.958 | 0.928 | 0.957 | 0.00099 | 0.071 | 8 | 14 | 10–2–10–4 (0.0007) |
3:5:1 | 0.923 | 0.956 | 0.914 | 0.965 | 0.0013 | 0.086 | 1 | 6 | 10–2–10–4 (0.0004) |
3:4:1 | 0.908 | 0.877 | 0.920 | 0.802 | 0.002 | 0.23 | 5 | 11 | 10–2–10–4 (0.0003) |