Forecasting Air Passenger Demand for Transportation Planning in Tourism Destination: A Comparative Analysis of Econometric and Time Series Models

Abstract

Accurate air passenger demand forecasting is essential for airport capacity planning and tourism transport policy, especially in destinations with strong seasonal travel patterns. This study compares econometric and time-series models for forecasting passenger demand at I Gusti Ngurah Rai International Airport, Bali. Monthly secondary data were used to represent passenger volume, tourist arrivals, hotel occupancy rate, gross regional domestic product, and a ticket-fare proxy. Multiple linear regression was applied as the econometric approach, while Seasonal Autoregressive Integrated Moving Average (SARIMA) was used as the time-series approach. Candidate models were evaluated using statistical feasibility tests, residual diagnostics, Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE). The best econometric model included tourist arrivals, hotel occupancy rate, and GRDP, with an adjusted R-squared value of 0.929 and statistically significant variables. The selected time-series model was SARIMA (0,1,1) (0,1,1)12. Out-of-sample validation using 2025 monthly data shows that SARIMA produced lower forecasting errors, with an MAE of 55,332, an RMSE of 63,049, and a MAPE of 5.8204%, compared with the econometric model’s MAPE of 12.7812%. The findings indicate that SARIMA is more suitable for operational forecasting, while econometric modeling remains valuable for explaining tourism and economic demand drivers.