This study investigates the effectiveness of integrating fuzzy logic with Recurrent Neural Networks
(RNNs) for forecasting financial time series. Traditional RNNs, while adept at capturing temporal
dependencies in sequential data, often struggle with the inherent uncertainty and variability present in
financial markets. To address this, we propose a fuzzy-based RNN model that combines the strengths of
fuzzy logic and RNNs. Fuzzy logic enhances the model's ability to manage imprecision and ambiguity
through fuzzy membership functions, rules, and inference systems. The experimental results demonstrate
that the fuzzy-based RNN outperforms standard RNN models in key performance metrics. Specifically,
the fuzzy-based RNN achieves lower Mean Absolute Error (MAE), Root Mean Squared Error (RMSE),
and Mean Absolute Percentage Error (MAPE), indicating improved forecasting accuracy and reliability.
This research highlights the advantages of incorporating fuzzy logic into RNNs, offering a more robust
approach to financial time series forecasting that can better handle the complexities and uncertainties
of financial data.
