Nowcasting the Number of Airplane Passengers at Ngurah Rai Airport Using Google Trends Data

Abstract

Data on the number of aircraft passengers is essential to airport managers and the government's policies. The policy relates to improving the facilities and capacity of airports and other affected sectors, such as the transportation and tourism industries. A policy taken will be better if the data used is very close to the time of policy decision-making. Therefore, a technique is needed to forecast very close to the current condition of the number of aircraft passengers, namely nowcasting. One of the data sources that can be used for nowcasting is Google Trends data. In this study, the identification of relevant keywords used for nowcasting, the formation of nowcasting models, and the search for the best model for nowcasting the number of aircraft passengers was carried out. The nowcasting methods used are SARIMAX and multilayer perceptron. In this study, five relevant keywords were generated for domestic departures and two for international departures. In the nowcasting modeling, the best model for nowcasting domestic departures is produced, namely the multilayer perceptron with MAPE and MAE values of 11.194% and 28.048 respectively, while for departures Internationally, the best model was produced, namely SARIMAX with MAPE and MAE values of 8,641% and 50,205 respectively.