<![CDATA[Stock price prediction is a complex problem in the financial domain due to the non-linear, dynamic nature of the data and its dependence on various external factors. This study proposes a deep learning–based approach using a Temporal Convolutional Network (TCN) to predict the stock price of Bank Central Asia Tbk. The model is evaluated under two scenarios: using a single feature (close price) and using three features MA5, and MA10. The dataset consists of historical BBCA stock data from 2010 to 2018, split into 80% training data and 20% testing data. The model is trained with 100 epochs, a window size of 60, batch size of 32, the Adam optimizer, and a learning rate of 0.0005. Experimental results show that the TCN model using a single feature achieves an RMSE of 227.1920, MAE of 192.8089, and MAPE of 4.3411%. Meanwhile, the TCN model with additional features (MA5 and MA10) demonstrates improved performance, achieving an RMSE of 176.4599, MAE of 145.7689, and MAPE of 3.2750%, indicating an accuracy improvement of more than 20%.These findings indicate that TCN is effective in capturing temporal patterns in financial time series data, and that incorporating simple technical indicators such as Moving Averages can significantly enhance model performance. This study contributes to the development of efficient and practical stock price prediction methods, particularly in the Indonesian stock market context.]]>
Copyrights © 2026
