<![CDATA[The volatile nature of stock price movements necessitates a prediction method capable of accurately modeling time-series data patterns. This study aims to predict the stock price of Bank Negara Indonesia (Persero) Tbk (BNI) using the Gated Recurrent Unit (GRU) method, based on historical data from the 2020–2025 period. The dataset includes closing prices, trading volume, moving averages (MA_7 and MA_21), volatility, and high-low ratios. The research process encompasses preprocessing, feature engineering, Min-Max normalization, the creation of sequential data with a time step of 30, and the splitting of data into training and testing sets. The GRU model is trained using the Adam optimizer, incorporating ReduceLROnPlateau and Early Stopping mechanisms to enhance training stability and mitigate the risk of overfitting. Model performance is evaluated using MSE, RMSE, MAE, R², and MAPE metrics. The results demonstrate that the GRU model yields effective predictions, achieving a MAPE of 3.91%, an RMSE of 211.21, and an R² of 0.746. Stock price predictions for the upcoming 10 days indicate a trend of gradual price decline. These findings suggest that the GRU method is effective for predicting BNI stock prices based on historical data and holds potential for supporting investment decision-making, while acknowledging external factors that influence market conditions.]]>
Copyrights © 2026
