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

EURASIP Journal on Advances in Signal Processing

Table 5 Performance comparison of various models and other models on the internet 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	√							232.3171	156.1111	0.9085
SVR	√							212.9878	130.6113	0.9245
LSTM	√							237.0122	141.6667	0.907
DenseNet	√							199.8171	122.5536	0.9362
STMNet	√					√		187.8049	116.6111	0.9448
STNet	√				√	√		186.0366	105.6131	0.9489
STCNet	√				√	√	√	167.9386	96.0912	0.9532
att-MCSTCNet (Conv-LSTM)	√	√	√	√	√	√	√	132.0883	85.244	0.9711
att-MCSTCNet (Conv-GRU)	√	√	√	√	√	√	√	129.1509	84.8507	0.9723