Air Passenger Estimation Using Gravity Model and Learning Approaches: Case Study of Thailand

Abstract

Air passenger estimation is essential since air-travel demand continuously grows. This work proposes to derive an air-passenger estimation model using three forms of gravity model and two machine learning approaches, regression and neural network. Data used in this work are Thailand’s domestic air-passengers and affecting factors on air-travel demand collected from publicly available sources. The results show that both regression and neural network with one hidden layer provide low error. Gross domestic product and number of tourists change in the same direction with air-passenger demand. The outcomes of this work give more understandings in employing machine learning to estimate air passengers in Thailand and can be developed for more complex forecast models in the future.