<![CDATA[Red bird’s eye chili is one of the food commodities widely needed by the community and tends to experience price fluctuations. This situation can create difficulties for consumers, traders, and food business operators in planning purchases. This is due to rapid price fluctuations, which may delays the chili purchases while prices remain unstable. In Surabaya, high demand and limited local production contribute to price variability. This study aims to predict red bird’s eye chili prices in Surabaya for a short-term horizon of 14 days by applying a hybrid ARIMA-LightGBM model optimized with Optuna. The data used in this study are historical red bird’s eye chili price data from Genteng Market and Keputran Market. ARIMA is used to capture linear patterns in time series data, while LightGBM is applied to model the residual patterns that cannot be fully explained by ARIMA. Optuna optimization is used to obtain optimal hyperparameter configuration. The evaluation results showed that the Optuna-optimized ARIMA-LightGBM model achieved the lowest forecasting errors among the evaluated models. The model obtained a MAPE of 22.84% for Genteng Market and 23.55% for Keputran Market. The comparison with Seasonal Naive, ARIMA, and ARIMA-LightGBM without optimization demonstrated that the hybrid ARIMA-LightGBM model with Optuna optimization provides improved prediction performance for short term red bird’s eye chili price prediction.]]>
Copyrights © 2026
