Research on passenger flow forecasting model for urban rail transport

Abstract

The trend and periodicity in metro passenger flow series data can provide some predictive power. Using AFC data, correlation analysis is used to verify the degree of correlation between current and historical passenger flow data. Through analysis of the original sample data and extraction of passenger flow features, Random Forest (RF), Long Short-Term Memory Networks (LSTM) and extreme gradient boosting (XGBoost) can be constructed respectively. The prediction model was used to forecast the future day rail passenger flow, and the accuracy of the model was evaluated by the mean absolute error, root mean square error and coefficient of determination. The model prediction and evaluation results show that the XGBoost model performs better, and the errors of the evaluation indicators are smaller than those of other models, and the algorithm is more accurate.