<![CDATA[The tourism sector makes an important contribution to supporting regional economic growth. Among the various provinces in Indonesia, West Nusa Tenggara (NTB) stands out as one of the main tourist destinations that has shown a fairly rapid increase in the number of tourist visits in recent years. This study uses Adam optimization and gradient clipping techniques to predict domestic and foreign tourist visits in NTB using the Long Short-Term Memory (LSTM) method. Monthly historical data for the period 2014–2023 from the NTB Tourism Office was processed through Min-Max Scaling normalization and divided with a ratio of 70:30 and 80:20. The LSTM model with a 4-layer architecture (2 LSTM layers with 50 units and 2 Dense layers) was tested using the Root Mean Squared Error (RMSE) metric. Based on the results obtained, the best configuration was shown at a ratio of 70:30 with 200 epochs, producing the lowest RMSE of 66.70 on the training data and 33.24 on the testing data. This implies that the model can capture seasonal patterns and visit trends, although it is less responsive to outliers such as natural disasters. This implementation provides a basis for tourism capacity planning and data-based destination management.]]>
Copyrights © 2025
