Performance Evaluation of LSTM, SVR, and Hybrid LSTM-SVR Models in Corn Commodity Price Prediction

Indonesia, a tropical country with abundant food resources, faces challenges in stabilizing essential food supplies due to price fluctuations influenced by crop failures. One way to help address these fluctuating prices is by using Machine Learning to accurately and effectively predict prices. This study explores the application of Deep Learning and Machine Learning models—Long Short-Term Memory (LSTM), Support Vector Regression (SVR), and a Hybrid LSTM-SVR approach—to forecast national average corn prices. Bayesian Optimization is employed for hyperparameter tuning, alongside data preprocessing and algorithm integration. The SVR model demonstrates the lowest Mean Absolute Error (MAE) of 38.76, followed by the LSTM model with 47.29, while the Hybrid LSTM-SVR model shows potential for improvement with an MAE of 252.47. Despite performance variations, all models effectively capture price trends, highlighting the need for further refinement to enhance predictive accuracy.