Perbandingan Metode Long Short-Term Memory (LSTM) dan Gated Recurrent Unit (GRU) dalam Memprediksi Harga Saham Telkom

Abstract

Accurate stock price prediction remains a challenging task due to the highly volatile nature of financial markets and the influence of various macroeconomic factors and market sentiment. PT Telkom Indonesia Tbk (TLKM), one of the largest publicly listed companies in Indonesia, has attracted significant attention from investors because of its substantial market capitalization and active stock trading. This study aims to compare the performance of the Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) models in predicting TLKM stock prices using time series data. The dataset consists of historical TLKM stock data, including the Open, High, Low, Close, Adjusted Close, and Volume variables. Data preprocessing involved data cleaning, normalization using the Min-Max Scaling technique, and time series sequence generation through the sliding window approach. Both LSTM and GRU models were developed using comparable network architectures and trained with the Adam optimizer and the Mean Squared Error (MSE) loss function. Model performance was evaluated using Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). The experimental results demonstrate that both models effectively capture historical stock price patterns. However, the GRU model consistently outperformed the LSTM model by achieving lower prediction errors while requiring lower computational complexity and training time. These findings suggest that GRU is a more effective and computationally efficient approach for predicting TLKM stock prices based on time series data.