Optimizing Factors for Accuracy of Forecasting Models in Food Processing Industry: A Context of Cacao Manufacturers in Vietnam

摘要

Manufacturing activities, in general, consume nearly 35% of total global production of electricity and are responsible for nearly 20% of total global carbon emissions (Graedel et al., 2011). According to Industrial Development Report, the manufacturing sector contributes one in six jobs globally. Since the financial crisis of 2008, there has been increased debate on maintaining sustained growth (Dubey et al., 2015). Food manufacturers have been shifted into the position whereby they have to deal with recent trends of high and volatile commodity prices, transportation and energy cost. For example, “transportation systems are essential to sustenance of human life and business growth. At the same time, they are also source of several negative impacts on human life and their environment. Therefore, they should be effectively controlled to achieve the socio-economic environmental objectives of sustainability” (Sayyadi and Awasthi, 2018a). Consequently, the urge to gain competitive advantages while staying commited to the quality and healthy margins is getting stronger. Demand forecasting can be an advantage or a drawback to a company. Especially, when the products have short-life cycle, it is complicated for transportation, storage and quality management; therefore, an accuracy forecasting would schedule for production planning to avoid later obstacles. Based on the assessment of the financial and logistics information of the Puratos Grand-Place Indochina, five different forecasting techniques including ARIMA, Exponential smoothing, GM(1,1), DGM(1,1) and Verhulst are employed, and their results are evaluated. The results indicate that DGM(1,1) has the best performance with the smallest error. The second best methods were the GM(1,1) and Verhulst. This result strongly supports the claim that Grey Forecasting Models can deal with small, limited and violated sequences of data input. In addition, since the forecast values show small differences from the actual values; if proper investigation can be done on this matter, it would create a huge impact on the company performance for having an accurate prediction of future events.