Research Article

An Improved Deep Spatial-Temporal Hybrid Model for Bus Speed Prediction

Table 6

Comparison analysis of different methods on weekdays/weekends.

MethodsWeekdaysWeekends
MAEMAPE (%)RMSEMAEMAPE (%)RMSE

ARIMA4.287610.186.73254.14647.066.3364
MLP4.764011.156.53444.56897.036.1384
RBFNN5.438111.857.08235.14638.476.9952
TM-CNN3.81128.494.55433.76535.964.4355
FDL3.35297.354.49383.46526.844.6941
DNN-BTF3.18257.133.89272.95686.513.5088
DHSTN2.79256.953.49432.38986.052.9338