THAI COCONUT PRICE FORECASTING USING ARIMA MODEL

Coconut is one of Thailand's most economically important crops. Due to the coconut price crisis, this study will identify the important features of the time series on monthly coconut prices in Thailand from January 2013 to December 2018. Then select independent variables, that influence on the domestic coconut price. ARIMA models were used to model and predict coconut prices. The appropriate model for forecasting the coconut price of Thailand was the ARIMA (2,1,0) model, which could predict price variability up to 95.2% with MAPE 6.215%. Considering the coconut import variables, it was found that ARIMAX (1,1,0) was the appropriate model that could demonstrate coconut price volatility up to 96.1% and MAPE 5.796%. It was found that the model with independent variables could provide improved accuracy results compared to the model without independent variables by selecting relevant factors to generate the forecast model. The model cloud supports the government to establish or release measures to protect the coconut price in Thailand. Results showed that the previous 3-month observations and coconut imports were a significant coconut price factor. When there are amounts of coconut imports, they will affect the coconut price for the next three months.