Enhancing Air Cargo Demand Forecasting Using Quantitative Data and the Long Short-Term Memory (LSTM)

Air freight plays a pivotal role in the supply chain for high-value cargo, which contributes 35% of global trade by value with only 1% by volume. With the economic development and rise of e-commerce, the air freight industry is experiencing rapid growth. One major type of air cargo is perishable goods, which requires timely delivery. Besides, some suppliers of air cargo have made commitments to deliver goods on specific dates. Hence, the air freight delay or disruption could lead to significant economic losses and declining customer satisfaction. One of the causes of delay is inaccurate capacity planning, which leads to the carriers not fulfilling the air cargo demand. Hence, the utilization of a precise prediction model is a vital asset for improving the air cargo capacity planning procedure. An accurate prediction model is instrumental in elevating the service level of airport authorities. Because of the impact of numerous socio-economic factors and unforeseen events, it is still a challenge to capture the patterns from quantitative data of air cargo demand and obtain favorable results. In this research, we utilized the Los Angeles air cargo dataset, which includes the monthly air cargo weight from 2006 to 2023 as our case study. An efficient forecasting method based on Long Short-Term Memory (LSTM) networks has been developed to address the intricacies and variability in cargo demand. The results show the proposed model effectively captures trends and seasonal variations in the time series data, providing accurate predictions of air cargo demand. This study provides insights for improving cargo management and resource allocation by advancing air cargo demand forecasting.