Forecast analysis of railway passenger flow based on time series analysis and LSTM

This paper aims to employ time series analysis and the Long Short-Term Memory network model to accurately forecast and analyze railway passenger flow, thereby enhancing the efficiency of transportation planning and management. Initially, we collected historical monthly passenger flow data for railway transportation. Through time series preprocessing, we identified and addressed trends and missing values to ensure the accuracy of the model input. The LSTM network is capable of capturing long-term dependencies. During the model training process, we utilized cross-validation and hyperparameter tuning to enhance the model’s generalization ability and prediction accuracy. The experimental results demonstrate the LSTM model’s clear superiority in predicting railway passenger flow, with a significantly lower prediction error compared to traditional time series models such as ARIMA. Additionally, a sensitivity analysis was conducted to assess the impact of various factors on the forecast outcomes, providing valuable insights for policymakers.