Research on Cargo Forecasting Based on ARIMA Time Series Forecasting and LSTM Models

Abstract

With the prosperity of the domestic e-commerce market, the volume of express delivery business climbs dramatically, which poses an unprecedented challenge to the cargo volume prediction of the warehouse in the transit centre. This paper introduces traditional logistics sorting centre cargo volume prediction models, such as grey prediction model, Bayesian deep network generalized linear model, ARMA model, etc., as well as current common prediction methods including qualitative prediction, linear regression prediction, time series prediction and neural network prediction. Then, the improved ARIMA and LSTM models are proposed to predict the cargo volume of logistics sorting centres based on the improved ARIMA and LSTM models, and the construction and improvement methods of these two models are described in detail, including the SARIMAX model and the improved LSTM model based on the attention mechanism. Finally, this paper describes the construction method of the two combined models, and selects the most suitable prediction model through comparative analysis, and predicts the daily and hourly cargo volume in the next 30 days and carries out experiments and analyses.