SSA and ARIMA for Forecasting Number of Foreign Visitor Arrivals to Indonesia

Abstract

Singular Spectrum Analysis (SSA) is the technique of non-parametric analysis of time series used for forecasting. SSA aims to decompose the original time series into a summation of a small number of components that can be interpreted as the trend, oscillatory components, and noise. The purpose of this research is to understand how the SSA model in forecasting the number of foreign tourist arrivals to Indonesia through four entrances. The result of forecasting obtained by using SSA will be compared with ARIMA method to assess its superiority. The data used in this study are the data of the number of foreign tourist arrivals to Indonesia through four entrances in the period January 1996 to August 2016. Four entrances used in this study are Ngurah Rai Airport, Kualanamu Airport, Soekarno-Hatta Airport, and Juanda Airport. The level of forecasting accuracy generated by each forecasting method is measured using the Mean Absolute Percentage Error (MAPE) criterion. The results showed that SSA method is the best forecasting method for forecasting the number of foreign tourist arrivals through Ngurah Rai Airport with an average MAPE value of 9.6%. Forecasting the number of foreign tourist arrivals through Kualanamu Airport, ARIMA method is the best forecasting method with an average MAPE value of 22.4%. In forecasting the number of foreign tourist arrivals through Soekarno-Hatta Airport, ARIMA method is the best forecasting method with an average MAPE value of 10.5%. In forecasting the number of foreign tourist arrivals through Juanda Airport, ARIMA method is the best forecasting method with an average MAPE value of 9.9%.