Predicting airline passenger load : A Case Study

• Main Authors: MA, Nang Laik, CHOY, Murphy, Sen, Prabir
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2014
• Subjects: airline; passenger load; predictive model; decision tree; demand planning; forecasting; simulation; Computer Sciences
• Online Access: https://ink.library.smu.edu.sg/sis_research/2445
• Institution: Singapore Management University

Airline industry has been growing at an outstanding rate with an annual growth rate about 6% worldwide in passenger load for the past decade. Airport transport industry around the globe has faced extreme challenge of handling high volumes of passengers especially in Asia due to the economy growth and most of them are already operating at 80% - 90% of their capacity in the recent year. The major clients – the airlines for the airport are also facing enormous pressure from the passengers to provide good services starting from the airport. Currently, the airport is using a fixed passenger load of Y% for all the airlines to predict the passenger flows for departure flights. The airport operations departments are also using it a guide to do the resource planning such as check-in counters to be opened and manpower required to man these counters. However, from the initial analysis, we have identified that the passenger load varied largely due to factors such as type of airlines (Full cost carrier verse low cost carrier), aircraft type, destinations etc. In this paper, we are going to analyze the passenger load from the past historical pattern and develop a predictive model using decision tree (DT) to forecast the passenger load based on certain criteria. The model is being tested against the actual data given for a particular month and the root mean square error of 3%-12% is observed for all the airlines at the airport. It shows the usefulness of the model in the real-world to predict the passenger load which will be useful to do the resource planning at the airport for day to day planning. Finally, a simulation model has been developed using the predicted passenger load as an input to compute the optimal number of check-in counters required to meet the service level agreement.