Application of the Autoregressive Integrated Moving Average (ARIMA) Model for Forecasting Indonesia's Non-Oil and Gas Export Values

ABSTRACT: Forecasting non-oil and gas exports is essential for supporting economic planning and trade policy in Indonesia, as this sector represents the largest contributor to the country's export earnings. The dynamic nature of export values, influenced by changes in global demand, commodity prices, and macroeconomic conditions, necessitates an accurate forecasting approach. This study aims to develop an Autoregressive Integrated Moving Average (ARIMA) model for forecasting Indonesia's monthly non-oil and gas export values. Monthly export data from January 2012 to December 2024 were obtained from Statistics Indonesia (BPS). The dataset was divided into a training set (January 2012–August 2023) and a testing set (September 2023–December 2024). The modeling procedure included stationarity testing, parameter identification using the autocorrelation function (ACF) and partial autocorrelation function (PACF), parameter estimation, residual diagnostic checking, and model validation. Based on the Akaike Information Criterion (AIC), parameter significance, and diagnostic tests, the ARIMA (0,1,1) model was selected as the best forecasting model. The model produced a Mean Absolute Percentage Error (MAPE) of 7.0187% for the training data and 6.0948% for the testing data, indicating highly accurate forecasting performance. The relatively consistent MAPE values between the training and testing datasets demonstrate that the selected ARIMA model generalizes well to unseen data. Therefore, the ARIMA (0,1,1) model can be considered an effective statistical approach for short-term forecasting of Indonesia's non-oil and gas exports and may provide useful information for economic planning and export policy formulation. Keywords: ARIMA; time series forecasting; non-oil and gas exports; Indonesia; MAPE.