Table 10.
Forecast 1 and forecast 2 to predict humidity indoor (refrigerator).
| Validation | Forecast 1 − humidity | Forecast 2 − humidity |
|---|---|---|
| Accuracy (%) | 96.42 | 98.66 |
| Batch size | 32 | 32 |
| Optimizer | Adam | Adam |
| Activation | Relu | Relu |
| Root Mean Squared Error − RMSE | 0.055 | 0.058 |
| Coefficient of determination (r2 score) | 0.97 | 0.96 |
| Epochs | 200 | 100 |
| Learning rate | 0.005 | 0.005 |
| Dense layer at first level (neurons) | 150 | 20 |
| Dense layer at second level (neurons) | 50 | 10 |
| Input layer forecast 1 − humidity Input layer forecast 2 − humidity |
4 (curr1, temp2, hum1, act1) | 3 (temp2, hum1, act1) |
| Output layer (humidity indoor) | 1 | 1 |
| Peak RAM Usage (bytes) | 1.4 K | 1.2 K |
| Flash Usage (bytes) | 18.9 K | 10.5 K |