Forecasting tourist visits using switching-Markov ARX based on Google Trends data

Forecasting plays a crucial role in effective planning and strategy within the tourism sector. Its accuracy relies on the availability of rich and diverse data in the digital age, which can be gathered from various online sources such as websites, search engines, and social media. However, tourism data can undergo unexpected shifts due to factors such as political events, societal changes, and health-related emergencies. The identification of structural data changes through Chow testing in tourism data underscores the need for precise forecasting using the Switching-Markov model. This research employs an advanced method to enhance forecasting accuracy, known as the Switching-Markov Autoregressive Model with Exogenous Input (SMARX). This model is an improvement over the Switching-Markov Autoregressive (SMAR) model, incorporating exogenous variables that exert influence on the dependent variable. This dynamic framework adapts to structural shifts in the relationship between these exogenous factors and the dependent variables. The primary focus is on predicting domestic tourist visits to Bali, integrating the Google Trends index as a significant input. This innovative approach provides more precise insights by considering the impact of online search trends on tourist behavior within Bali's tourism industry. The SMARX model outperforms the SMAR model, demonstrating a lower Mean Absolute Percentage Error (MAPE) of 0.271 compared to 0.304. This indicates that the SMARX model offers greater predictive accuracy and reliability.