Forecasting and Research Based on LSTM Time Series Forecasting Algorithm and SARIMA Model

In this paper, for the difficult problem of cargo volume prediction in the sorting centre, the Long Short-Term Memory Network (LSTM) in deep learning and the classical statistical model Seasonal Autoregressive Integral Sliding Average Model (SARIMA) are used to conduct in-depth research. By constructing the corresponding prediction models and combining them with the actual business datasets, LSTM and SARIMA are applied to predict cargo volume and the prediction effects are analysed. Eventually, the thesis will propose method selection suggestions applicable to sorting centre volume forecasting based on the empirical results, as well as potential hybrid forecasting framework ideas, which will provide strong support for improving the operational efficiency and intelligence of logistics systems.