Adopting new ARIMA double layering technique in make-to-stock production policy

In supply chain management, the employed forecasting algorithm plays a very vital and crucial role in how a particular business performs well in the market. This study proposes a time series based forecasting model to help in designing an effective make-to-stock supply chain mechanism for tackling the large number of suppliers (suppliers diversion) and products (items diversion). The proposed technique extends the traditional ARIMA model to ARIMA Double Layering Technique (DLT) where the algorithm is implemented in two layers: first one is to obtain the number of orders level, and the other layer is implemented at the quantity level. The systematic implementation procedure of our approach is illustrated through a real case study. The accuracy evaluation and consistency of the results show that ARIMA-DLT provides superior forecasting when compared with ML-based models which provides insightful managerial views to other similar forecasting problems.