Forecasting Tourism Visitor Numbers Using a Recurrent Neural Network with a Long Short-Term Memory Algorithm

Accurate forecasting of visitor numbers is essential in tourism management to ensure service quality and visitor satisfaction, especially during peak seasons such as holidays and weekends. This study addresses the lack of a predictive tool at PT Taman Impian Jaya Ancol (TIJA), a major recreational destination in Indonesia, by developing a forecasting model for visitor numbers. The research utilized monthly time series data of visitor numbers from January 2012 to December 2022. A Deep Learning approach was applied using the Recurrent Neural Network (RNN) architecture with the Long Short-Term Memory (LSTM) algorithm. The dataset was split with an 80:20 ratio for training and testing, normalized using the RobustScaler technique, and optimized with the ADAM optimizer. The model achieved a minimum Mean Squared Error (MSE) of 0.3095 and a prediction accuracy of 94.85%. These results indicate that the LSTM model can effectively predict visitor trends. The findings are expected to support TIJA and other tourism operators in preparing resources and facilities in advance, improving operational planning, and enhancing the overall visitor experience.