<![CDATA[The Bantul Regency Tourism Office faces challenges in collecting tourist attraction data because it is still done manually. This study aims to develop a web-based prediction system using the Seasonal Autoregressive Integrated Moving Average (SARIMA) method. This system is designed to facilitate data collection on tourist attractions while generating predictions of visitor numbers. Based on the results of accuracy testing using MAE (Mean Absolute Error) and MAPE (Mean Absolute Percentage Error), the Parangtritis and Depok Beach tourist areas have MAE values of 35,157.41 and MAPE 22.75%, indicating a fairly large absolute prediction error but still reasonable considering the high volume of visits. Meanwhile, Samas Beach recorded the highest MAPE value of 165.22%, due to data fluctuations that make predictions inaccurate. Conversely, predictions for Goa Cemara Beach, Kwaru Beach, Goa Selarong Area, and Goa Cerme Area have MAPE values below 15%, indicating the model is able to provide fairly good prediction results with a small average error. However, at Pandansimo Beach, the MAPE value reached 46.47%, indicating the model was not yet adequate for this location due to unstable data trends. The results showed that the SARIMA model can be applied to a system to predict tourist visits, but with varying levels of accuracy at each tourist destination, depending on the stability of each tourist destination's data. Keywords: Tourist visit prediction, SARIMA, time series forecasting, web-based system, Bantul Regency Tourism Office]]>
Copyrights © 2025
