A Comparative Analysis of Forecasting Methods for Passenger Volume through Suvarnabhumi Airport

Abstract

This research aims to investigate and compare forecasting methodologies and determine the optimal forecast horizon for predicting passenger volume through Suvarnabhumi Airport. The study utilized daily domestic and international passenger data published by the Civil Aviation Authority of Thailand from March to December 2024. The dataset was partitioned into a training set (March–November 2024) for model development and a validation set (December 2024) for performance evaluation. Three forecasting methods were implemented: linear trend analysis, exponential smoothing, and time series decomposition. Model performance was evaluated using Mean Absolute Deviation (MAD) and Mean Absolute Percentage Error (MAPE) as accuracy metrics. Data analysis was conducted using Microsoft Excel software. Findings indicated that for domestic passenger forecasting, the exponential smoothing method with a smoothing parameter (α) of 0.8 and a one-week forecast horizon yielded optimal results. For international passenger forecasting, the exponential smoothing method with a smoothing parameter (α) of 1.0 and a one-month forecast horizon demonstrated superior predictive performance.