Table 4. Comparison of models with processing times and normalization applied.

Models	Time (min)	Normalization “scaler_type”:
TFT	18.49	robust, standard
RNN	17.08	robust
LSTM	42.49	robust
GRU	59.15	robust
TCN	2.34	robust
DeepAR	6.90	standard
DilatedRNN	3.40	robust
BiTCN	0.83	standard
VanillaTransformer	5.89	robust
Informer	6.26	robust
Former	8.98	robust
FEDformer	2.87	robust
PatchTST	2.03	robust
iTransformer	0.29	robust