Machine Learning Model of Seasonal Autoregressive Integrated Moving Average with Exogenous Variables for Logistics

The article considers the problem of forecasting the volume of seasonal logistics transportation of fruits and vegetables using time series based on the seasonal autoregressive integrated moving average with exogenous variables (SARIMAX) machine learning model. To analyze and build forecasts for a data set, the volumes of logistics transportation of fruits and vegetables are analyzed taking into account seasonality. The purpose of this article is to develop and study a model for forecasting the volume of seasonal logistics transportation using SARIMAX. The problem of time series analysis for determining seasonal fluctuations using the SARIMAX machine learning model is solved. The study shows the effectiveness of the constructed SARIMAX model for forecasting seasonal logistics transportation in comparison with other models, such as autoregressive integrated moving average (ARIMA), random forest model, exponential smoothing, Holt-Winters model, artificial neural network (ANN), multilayer perceptron (MLP), GARCH model, support vector model. The evaluation of the modeling results shows the effectiveness of the SARIMAX model with an R-square of 0.99 and a low mean absolute error (MAE) of 0.045. Python programs were developed to solve the problem of modeling and forecasting logistics transportation taking into account seasonality. The results show the accuracy of the forecasting model of about 96% and its effectiveness in applying to a wide range of forecasting problems.