Passenger Flow Prediction Using Smart Card Data from Connected Bus System Based on Interpretable XGBoost

Table 7

Comparison of the average value of the evaluation indicators of 30 bus stations.

Model	MAPE (%)	RMSE	MAE	Time (s)
The proposed model	42.90	4.76	3.53	10.63
XGBoost model without the number of buses arriving	55.86	5.46	4.08	—
KNN regression model	61.08	6.03	4.44	11.78
BP neural network model	50.85	5.70	4.31	40.21
LSTM model	44.11	6.17	4.63	28.15