Comparison Between Neural Network and Grey System Models for Cooking Oil Price

Abstract

Cooking oil is one of the primary raw materials used in Indonesia. In this study, a comparison of the two forecasting models from the two methods, namely Neural Network and Grey System, was carried out. Forecasting is carried out on cooking oil raw materials, namely CPO production volume and demand for related products, namely biodiesel, to analyze changes in cooking oil prices. The appropriate forecasting model is expected to be able to describe the pattern of cooking oil price fluctuations for the following few periods. The criteria for selecting the best model use the minimum MAPE testing value. The results show that the Grey System method produces the best forecasting model for biodiesel demand data with a small amount of data, while for the CPO variable, which has a larger amount of data, the best model is obtained using the Neural Network model, with the MLP (3-3-1) architecture.