FORECASTING THE NUMBER OF ASEAN TOURISTS TO INDONESIA WITH THE HYBRID SINGULAR SPECTRUM ANALYSIS-SUPPORT VECTOR MACHINE METHOD

Tourism plays a crucial role in Indonesia's economy, significantly contributing to the nation’s revenue, employment, and overall growth. ASEAN countries represent one of the largest sources of international tourists, making their travel patterns essential for Indonesia’s tourism sector. Accurately forecasting the number of tourists from ASEAN nations is crucial for formulating effective policies and strategies to enhance tourism development. This study aims to provide a reliable forecast of ASEAN tourist arrivals to Indonesia, utilizing the hybrid Singular Spectrum Analysis-Support Vector Machine (SSA-SVM) method. Monthly time-series data spanning from January 2017 to June 2024 is analyzed to predict future trends in tourist visits. The data, sourced from Badan Pusat Statistik (BPS), captures detailed records of ASEAN tourist arrivals. SSA is applied to decompose the data into key components such as trend, seasonality, and noise, which are then processed by the SVM model to forecast future arrivals. This hybrid approach effectively handles non-stationary data and captures intricate patterns that are often missed by traditional forecasting techniques. Unlike conventional methods, SSA-SVM offers a more robust solution to complex time-series forecasting, as it combines the power of SSA in data decomposition with the precision of SVM in pattern recognition. The model’s accuracy is tested against ARIMA, Exponential Smoothing, and standalone SVM, with results showing superior performance. The SSA-SVM model yields the lowest error rates, achieving MAPE (0.220%), MAE (33,274.64), and RMSE (43,716.18), outperforming the other models. These results indicate the model's higher precision and reliability in predicting ASEAN tourist arrivals. By offering better generalization and reducing forecasting errors, SSA-SVM provides valuable insights for tourism stakeholders and policymakers, helping to allocate resources more efficiently, design targeted marketing strategies, for fluctuations in tourism demand. The findings also highlight the potential of SSA-SVM for improving the accuracy of long-term tourism forecasts, ultimately contributing to sustainable tourism development in Indonesia. This research opens avenues for further exploration into machine learning and hybrid forecasting methods in tourism, aiming for more adaptive, data-driven decision-making in the sector.