Application of the Lee Fuzzy Time Series Method in Forecasting Retail Rice Prices in West Sumatra

Abstract

Retail rice prices are one of the important food commodities that affect economic stability in West Sumatra Province. Fluctuations in rice prices occurring in each period require forecasting to support decision-making related to food policies. This study aims to forecast retail rice prices in West Sumatra using the Lee Fuzzy Time Series method and to measure the forecasting accuracy. One of the advantages of the Lee Fuzzy Time Series (FTS) method is that it does not require stationarity assumption testing. The Lee Fuzzy Time Series method is designed for short-term forecasting and can be applied to both stationary and non-stationary data. In addition, this method can handle uncertainty and fluctuations in the data. Therefore, the Lee Fuzzy Time Series method is suitable for forecasting rice prices, as rice price data are classified as time series data. The data used were monthly retail rice price data from January 2020 to December 2024 obtained from the Central Statistics Agency (BPS). The forecasting stages included determining the universe of discourse, interval formation, fuzzification, establishing Fuzzy Logical Relationships (FLR), forming Fuzzy Logical Relationship Groups (FLRG), and defuzzification. Forecasting accuracy was measured using Mean Absolute Percentage Error (MAPE). The results showed that the Lee Fuzzy Time Series method produced a MAPE value of 1.27%, which is categorized as very good. The forecasting result for January 2025 indicated that retail rice prices were predicted to increase compared to the previous period.