Forecasting the Palm Oil Market: A Comparative Study of LSTM and Bi-LSTM Models for Price Prediction

Pieter, Franky Bryan
Suharjito

Jurnal Sains Dan Teknologi (SAINTEKBU)
Vol. 16 No. 02 (2024): Vol. 16 No. 02 August 2024

This study underscores the critical need for accurate palm oil price predictions amid market volatility, driven by factors like demand shifts and supply disruptions. Employing advanced neural network models, specifically Long Short-Term Memory (LSTM) and Bidirectional LSTM (Bi-LSTM), the research spans May 2007 to December 2022 using Market Insider data. Evaluation metrics, including RMSE 0.000083 and MAPE 0.76%, highlight Bi-LSTM's superior predictive prowess. Beyond immediate benefits for decision-making, the study emphasizes broader impacts on market stability, reducing volatility and fostering sustainability in the palm oil industry. Overall, this paper showcases the efficacy of Bi-LSTM in enhancing palm oil price prediction accuracy, offering practical insights, and contributing to industry sustainability.