Subway Passenger Flow Prediction Based on XGBOOST with Weather Factors

Abstract

Since the beginning of this century, subways have increasingly occupied a higher share of daily transportation in Chinese cities. The growing network traffic has imposed higher demands on operating companies. Therefore, accurate passenger flow prediction is crucial for operational management and precise service level matching. With the maturity of machine learning methods, there are numerous cases of using these methods for passenger flow prediction. However, the influence of weather factors has been underexplored, despite its impact on passenger travel. This paper uses an XGBoost training model to obtain data on subway passenger flow, weather, maximum temperature, and minimum temperature. It examines the influence of day of the week, weather, and temperature on subway passenger flow, while also scoring the importance of these factors. The results show that the minimum temperature is the most influential factor in the model. The final model's fit is indicated by an R-squared value of 0.79, which can achieve a certain degree of accuracy in passenger flow prediction.