Improving Food Supply Chain using Hybrid Semiparametric Regression Model

Abstract

A demand variability in food retail sector affects production, ordering, and purchasing decisions in the entire upstream food supply chain, which in turn result in food waste and stock-outs. The volatility in demand for perishable fresh foods mainly occurs due to the demand influencing factors such as seasonality, temporary price reductions, holidays, and festivals. In particular, own- and cross-price deal effects between prod-ucts are some of the important causes of bullwhip effect in the food supply chain. Therefore, it is necessary to develop a forecasting model which considers all the de-mand influencing factors in a proper way to improve the forecast accuracy. The main objectives of this study is (i) to improve the standard semiparametric regression (SR) model into a hybrid auto regressive integrated moving average – semi parametric regression (ARIMA-SR) model and (ii) to assess the price deal effects. For the purpose of investigation, the daily sales data of perishable fresh foods from a retail store in Germany is used. From the obtained results, it has been identified that the ARIMA-SR model has high adjusted R2 and low forecast error, when compare to the existing traditional models.