A COMPARISON OF TIME SERIES MODELS FOR FORECASTING THE NUMBER OF THAI TOURISTS VISITING NAKHON SI THAMMARAT PROVINCE

Abstract

This study aims to 1) Compare the performance of time series models for forecasting the number of Thai tourists visiting Nakhon Si Thammarat Province, and 2) Forecast the number of Thai tourists visiting the province over a 24-month horizon. The study employs a quantitative research approach using monthly secondary data obtained from the Ministry of Tourism and Sports, covering the period from January 2020 to December 2025. Time series analysis techniques were applied, and two forecasting models were compared: Holt–Winters Exponential Smoothing and the Seasonal Autoregressive Integrated Moving Average (SARIMA) model. The exploratory data analysis revealed that the number of Thai tourists exhibits typical time series characteristics, including trend, seasonality, and high volatility. These features were particularly pronounced during the COVID-19 pandemic, which introduced non-stationarity and structural changes in the data during certain periods. Forecasting accuracy was evaluated using the Root Mean Squared Error (RMSE) and the Mean Absolute Percentage Error (MAPE). The results indicate that the SARIMA model produced substantially lower RMSE values than the Holt–Winters Exponential Smoothing method. The findings suggest that the SARIMA model is more suitable than the Holt–Winters approach in contexts characterized by high volatility and structural disruptions. Beyond supporting tourism planning and policy formulation at the regional level, this study also provides methodological insights into the selection of appropriate time series forecasting models under conditions of high uncertainty.