Table 3.

Same as Table 2 but for the results of the fully connected DNN model predictions.

Fully connected deep neural network model statistics
Prediction size (forecast)	Train RMSE [°C]	Test RMSE [°C]	Train r2	Test r2	Train F1	Test F1
6 h	1.55	1.53	0.99	0.98	0.96	0.96
12 h	1.72	1.84	0.98	0.98	0.96	0.95
24 h	2.06	2.23	0.97	0.96	0.95	0.94
30 h	2.34	2.69	0.96	0.95	0.94	0.92
36 h	2.53	2.96	0.96	0.95	0.94	0.91
48 h	2.7	2.94	0.95	0.94	0.94	0.93

The DNN model uses a sequential model with elu activation.