Forecasting demand for products in distribution networks using R software

Purpose: This article addresses the issue of forecasting demand for products flowing in a distribution network conducted from the perspective of a 3PL logistics operator. Its purpose is to present a tool based on the R software, which is to be used for automatic forecasting of time series. Design/methodology/approach: The work uses the algorithm of automatic forecasting of time series implemented in the "forecast" package. The algorithm was used in a loop for different lengths of the time series to determine the best length of the series. The minimal RMSE value in the training set and in the test set were considered as optimality criteria. Findings: It is shown that the best time series length is 60 weeks in the considered case. Originality/value: The procedure for selecting the best time window length for forecasting demand for products in distribution network.

---

Opracowanie rekordu ze środków MNiSW, umowa Nr 461252 w ramach programu "Społeczna odpowiedzialność nauki" - moduł: Popularyzacja nauki i promocja sportu (2020).