Attention based multi-component spatiotemporal cross-domain neural network model for wireless cellular network traffic prediction

EURASIP Journal on Advances in Signal Processing

Table 3 Performance comparison of various models and other models on the Sms dataset

Model	Multiple factors							RMSE	MAE	R²
Model	F₀	F_r	F_d	F_w	F_m	F_s	F_c	RMSE	MAE	R²
LR	√							94.1253	35.6864	0.5765
SVR	√							72.6544	30.5419	0.7571
LSTM	√							69.0916	40.1357	0.7715
DenseNet	√							62.7586	27.512	0.8057
STMNet	√					√		58.6207	29.6203	0.8191
STNet	√				√	√		57.8182	28.132	0.8276
STCNet	√				√	√	√	52.8665	28.1034	0.8559
att-MCSTCNet (Conv-LSTM)	√	√	√	√	√	√	√	39.1589	22.6542	0.9274
att-MCSTCNet (Conv-GRU)	√	√	√	√	√	√	√	38.3087	21.8829	0.9303