<![CDATA[Stock price prediction presents a significant yet intricate challenge in financial forecasting, primarily due to volatile market dynamics and the nonlinear nature of data. This study investigates the efficacy of the Long Short-Term Memory (LSTM) model, a specialized Recurrent Neural Network (RNN), for forecasting the stock price of PT. Malindo Feedmill, Tbk., a publicly listed agribusiness firm on the Indonesia Stock Exchange. A five-year historical dataset of daily stock prices (open, high, low, close, volume) was utilized. Pre-processing involved data normalization, the application of a sliding window approach, and partitioning the data into training and testing subsets. The LSTM model was trained on sequential closing prices to effectively learn and model long-term dependencies inherent in stock price movements. The model's predictive performance was rigorously assessed using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE) metrics. Our results reveal that the LSTM model adeptly captures price trends, yielding a low MAPE of 3.47% on the test set. Comparative analysis against traditional models like linear regression confirms that LSTM provides superior accuracy and robustness, especially under volatile market conditions. This research highlights the significant potential of deep learning models in facilitating smarter investment decisions within the Indonesian agricultural sector. Subsequent work will aim to integrate sentiment analysis and macroeconomic indicators to further improve real-time predictive accuracy.]]>
Copyrights © 2025
