Analisis Perbandingan Model Double Exponential Smoothing dan ARIMA untuk Prediksi Harga Beras di Indonesia

Abstract

Rice prices in Indonesia tend to increase from year to year and are influenced by various factors, such as domestic production and seasonal factors. Therefore, rice price forecasting is an essential thing to do. This study aims to analyze and compare the performance of two forecasting models, namely Holt’s Double Exponential Smoothing (DES) and Autoregressive Integrated Moving Average (ARIMA), in predicting rice prices in Indonesia. The data used is secondary data of average wholesale rice prices from January 2021 to December 2024. The results show that the optimal parameters of the Holt’s DES model are alpha = 0,9 and beta = 0,1, while the best ARIMA model is ARIMA(2,2,1) . Both models have a high level of accuracy with a Mean Absolute Percentage Error (MAPE) value of less than 1%. However, the ARIMA(2,2,1) model has a lower MAPE value than Holt’s DES model. Hence, it is more accurate in modeling rice prices in Indonesia. The forecasting results show that Holt’s DES model tends to produce higher rice predictions than . This occurs because Holt’s DES model is more sensitive to increasing trends, while ARIMA tends to be more conservative in capturing patterns of price changes. Thus, the selection of a model for rice price forecasting should consider the characteristics of the trend that occurs in the market, whether it is experiencing a continuous increase or has a fluctuating pattern.