Forecasting corn prices in agribusiness using Causal Forecasting models

The objective of this research is to compare models for forecasting corn prices and study various factors from forecasting models that affect corn prices. The retail price of fertilizer crude oil price amount of maize production Northern rainfall the amount of maize imports export volume of corn for animal feed currency exchange rate yield of cassava and the price of cassava. The data used is the corn prices (baht/kilogram) monthly for 5 years and 9 months (January 2017 to September 2022) of the Office of Agricultural Economics (OAE). The data are divided into 2 sets; the first set, 60 months (January 2017 to December 2021) is to construct forecasting models, with Support Vector Machine, Neural Network and Dynamic Regression Model. And the second set, 9 months (January 2022 to September 2022) for forecasting testing with the Mean Absolute Percent Error (MAPE).

The results showed that the best model is the Support Vector Regression method. Because the average absolute error percentage is the smallest (4.8915) and the top 3 factors, that affect corn price, are the price of cassava, currency exchange rates and fertilizer prices.