<![CDATA[Stock price prediction is one of the main challenges in financial market analysis. This study develops a Deep Neural Network (DNN) model using TensorFlow to predict the stock price of PT Bank Central Asia Tbk (BBCA) based on historical stock price data, the Jakarta Composite Index (IHSG), and the USD/IDR exchange rate. The model is optimized through hyperparameter tuning using KerasTuner with the Hyperband method, allowing for more efficient exploration of hyperparameter combinations. The tuning process yielded the best model configuration with a learning rate of 0.001269, four hidden layers, and the application of Batch Normalization, L2 kernel regularization, and dropout in each hidden layer.The model was evaluated using various batch sizes (4, 16, 32, 64, and 128) with Mean Squared Error (MSE) and Mean Absolute Error (MAE) as performance metrics. The results show that batch size 128 provides the best performance, with MSE of 0.0094 and MAE of 0.0490 on the test data, indicating high accuracy and good generalization on unseen data. The best model was then implemented as an API using Flask, deployed to Google Cloud Run, and integrated with a Flutter-based mobile application. Testing confirmed that the API can handle requests quickly using TensorFlow and produce accurate predictions.Thus, this study successfully developed a DNN-based BBCA stock price prediction system that can be applied to data-driven investment decision-making.]]>
Copyrights © 2025
