Comparison Between SARIMA Model and Artificial Neural Network On Forecasting Foreign Tourist in Batam City

Abstract

Batam City is one of the tourist attractions in Indonesia with the number of foreign tourist arrivals increasing every year. As one of the impacts of increasing the number of foreign tourist visits, the provincial government must improve the existing facilities in the tourism area, both in quality and quantity. In order for these facilities to be adequate to serve foreign tourists visiting Batam City in the future, it is estimated that the number of tourist visits to Batam City in the future is expected. This study aims to model foreign tourist arrivals using the SARIMA method and Neural Networks and compare the accuracy of the two methods with Mean Squared Error (MSE) and Mean Absolute Percentage Error (MAPE). The best SARIMA model for data on the number of foreign tourist arrivals to Batam City is SARIMA (2, 1, 0)(1, 1, 0)12 with MSE = 2,672,774,359 and MAPE = 21,4487%. The Neural Network Model is ˆy = max(0, 0.03208266 + 0.48310924V1 +...+ 0.46732363V8) with MSE = 171.279.990 and MAPE = 7.1404%. Thus, modeling with Artificial Neural Networks in these cases provides a better model than SARIMA in modeling data on the number of tourist visits to Batam City.