Comparison between Linear Regression, Support Vector Machines, and Neural Networks in Forecasting the Number of Foreign Tourist Visits to Indonesia

Abstract

The tourism sector is one of the largest foreign exchange contributors to the country's economic growth. One of the determining factors for tourism development is the number of foreign tourist arrivals. It is necessary to carry out development, planning strategies that appropriate, effective, and efficient to increase foreign tourist arrivals. Business Intelligence is one of the digital trends and the main digital marketing strategy for Digital Tourism by implementing data mining for data analysis, including forecasting. There are various kinds of forecasting methods, but the problem is in choosing the best method for the problem and the desired output. This study aims to compare the accuracy between Linear Regression, Support Vector Machine (SVM), and Neural Network methods to forecast the number of foreign tourist arrivals to Indonesia using RapidMiner Studio and parameter of Root Mean Square Error (RMSE) to compare the accuracy of each method. The results of this study indicate the RMSE value of the SVM has the smallest RMSE value compared to the other two methods. SVM is used to make a forecast for the next period which can be used as a basis for the government and the tourism industry in making decisions.