<![CDATA[Exchange rates are a crucial indicator of a nation's economic stability. Fluctuations in exchange rates can influence investment decisions, international trade, and monetary policies. The prediction of foreign exchange rates against the Indonesian Rupiah holds significant importance in financial analysis and decision-making. Exchange rates are a crucial indicator of a nation's economic stability, influencing investment decisions, international trade, and monetary policies. However, accurately predicting foreign exchange rates remains a challenging problem due to the highly. This research aims to enhance prediction accuracy by proposing a CNN Bi-LSTM model with the addition of an Attention Mechanism (CNN Bi-LSTM AM). Historical exchange rate data were normalized and divided into training, testing, and validation sets. The CNN model extracts local features, BiLSTM captures bidirectional temporal patterns, and the Attention Mechanism highlights critical features to improve data processing efficiency. The evaluation was conducted using Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) metrics. The results show that CNN Bi-LSTM AM outperformed BiLSTM and CNN Bi-LSTM, achieving MAPE values of 0.7795%, 0.3877%, 0.4062%, 0.7646%, 0.4267%, and RMSE values of 0.1953, 0.2199, 0.2069, 0.1758, 0.3250 for USD, Pounds, EURO, Yen, and Franc, respectively. The addition of the Attention Mechanism significantly contributed to improved accuracy, distinguishing this model from previous research. This study supports the relevance of innovation in hybrid architectures and Attention Mechanisms for financial time series predictions]]>
Copyrights © 2025
