Research on Cargo Volume Prediction in Logistics Sorting Centers Based on ARIMA-LSTM Forecasting Algorithm

This paper presents a study on the prediction of cargo volumes in 57 logistics sorting centers using the ARIMA-LSTM forecasting algorithm. The primary objective is to forecast the volume of goods for the next thirty days as well as hourly predictions. The suitability of the ARIMA forecasting model is initially assessed by employing q-q plots and the Kolmogorov-Smirnov test to determine the data distribution. Subsequently, the LSTM forecasting model is integrated to enhance the accuracy of predictions. Following the prediction process, the presence of errors in the generated forecasts is evaluated through visual analysis. Any identified errors are then subject to fuzzy correction to improve the overall precision of the predictions.