SARIMA model for forecasting foreign tourists at the Kualanamu International Airport

摘要

Seasional Autoregressive Integrated Moving Average (SARIMA) model was built to predict the number of tourists arrived via the Kualanamu International Airport in Medan, Indonesia. The data size was 72 periods, taken from January 2008 to December 2013 for determining models, and 23 periods from January 2014 until November 2015 for testing the model accuracy. Steps of SARIMA involve describing the data characteristics, identification of model types, estimation of model parameters, diagnostics of parameter significance, selection for getting the best model, and forecasting for next periods. Based on the testing, data have to be transformed with certain values for non-seasonal differencing, seasonal differencing, and length of season. Furthermore, types of models were obtained through ACF as well as PACF plots followed by parameter estimations using Ordinary Least Square (OLS), and diagnostic steps for testing the parameter significance. The best model from the selection was SARIMA(1, 1, 1)(1, 1, 1) with twelve seasons.