Performance Analysis of SVM and Linear Regression for Predicting Tourist Visits in North Sumatera

Abstract

Indonesia, an archipelago rich in cultural diversity, historical heritage, and stunning natural scenery, offers an extraordinary travel experience to visitors who make this country their vacation destination. Tourism in Indonesia plays an essential role in the domestic economy, contributing to Gross Domestic Product. With its abundant natural and cultural resources, North Sumatra has long been recognized as an attractive destination for foreign tourists. However, the tourism sector faces significant challenges related to fluctuations in the number of visits, mainly due to the impact of the COVID-19 pandemic, which has disrupted global travel patterns and caused considerable uncertainty in tourism forecasting. Therefore, predicting the number of tourist visits becomes crucial for effectively planning and managing tourist destinations. This research aims to compare the performance of two forecasting algorithms, SVM and linear regression, in predicting foreign tourist visits in North Sumatra using historical data from 2019 to 2023. The dataset was subjected to a preprocessing phase to ensure data cleanliness and consistency, focusing on key variables such as seasonal trends, external factors, and market dynamics. Both models were evaluated based on two commonly used accuracy metrics, MAPE and RMSE, to assess how well the models could predict actual tourist arrivals. The results of the study indicate that Linear Regression outperforms SVM in terms of prediction accuracy, with a MAPE of 42.40% and an RMSE of 6735.6, compared to SVM with a MAPE of 46.65% and an RMSE of 8020.42. These findings provide valuable insights for local government authorities and tourism industry stakeholders to enhance destination planning, resource allocation, and strategies to attract more foreign tourists in the post-pandemic era.