<![CDATA[Stock price prediction is a major challenge in the financial sector due to nonlinear factors and data uncertainty. This study aims to develop a predictive model by integrating fuzzy logic into deep learning algorithms to improve accuracy and robustness against noise. This is a quantitative experimental study using 1,000 daily historical stock price data of BBCA (Bank Central Asia), collected via web scraping from public sources. The data were analyzed using three types of neural networks: Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU), both before and after fuzzy integration. Fuzzification was applied to the price data to generate linguistic features, which were added as input to the neural network models. The models were evaluated using Train Cost, Test Cost, and the number of epochs, and a t-test was conducted to assess the statistical significance of performance differences. Our findings show that the LSTM model with fuzzy input achieved the best performance, with a Train Cost of 0.0002 and a Test Cost of 0.0052, and demonstrated superior capability in handling long-term dependencies. In contrast, RNN and GRU models showed decreased accuracy after fuzzy integration. The combining fuzzy and LSTM model shows promise for broader applications in time-series forecasting under uncertainty.]]>
Copyrights © 2025
