Implementasi Model Fuzzy Time Series Markov Chain Dalam Forecasting Ekspor Minyak Kelapa Sawit Indonesia

Abstract

Palm oil exports are one of the strategic commodities that contribute significantly to national foreign exchange; however, their values tend to fluctuate annually, thus requiring an accurate forecasting method to support decision-making in the export sector. This study aims to implement the Fuzzy Time Series Markov Chain model to forecast Indonesia’s total palm oil exports using secondary data from 2010 to 2024. The methodological stages include determining the universe of discourse, constructing intervals using Sturges’ rule, performing fuzzification, forming Fuzzy Logical Relationship (FLR) and Fuzzy Logical Relationship Group (FLRG), developing the Markov transition probability matrix, and conducting defuzzification and adjustment processes. The forecasting accuracy was evaluated using the Mean Absolute Percentage Error (MAPE) by comparing the predicted values with the actual data. The results indicate that the predicted total palm oil export for 2025 is 25,268,721 tons with a MAPE value of 4.16%. This value falls into the very good category since it is below 10%, indicating that the Fuzzy Time Series Markov Chain model is capable of producing accurate forecasting results and can be used as a consideration in decision-making related to Indonesia’s palm oil export policies in the future

Keywords: Fuzzy Time Series Markov Chain, forecasting, palm oil export, MAPE.