Predicting Cockpit Crew Absence During Assignment Schedule Planning Using Classification And Regression Trees Algorithmin A National Flag Carrier Airlines Of Indonesia
In an airlines company, disruption is a common thing to happen during operations. It’s significant for the airline's industries to forecast or predict the source of disruptions to reduce the cost of schedule recovery due to schedule delay or cancel. There are many factors of disruption in the airlines which causes schedule delay or cancel, such as the mechanical problem of aircraft (maintenance), weather condition, crew sickness, security, etc. In this research, it highlights the absenteeism of the pilot due to sickness. A supervised learning method is proposed to predict the sickness of cockpit crew based on data given on the previous period. The classification and regression tree/decision tree algorithm use the feature of the cockpit crew as predictor variable to predict the future absenteeism of the pilot. The real data in 2017 is used to train and test the accuracy of the model. The result shows that administrative or human resource and historical sickness data can be the predictor to build model for cockpit crew sickness prediction. In this research most pilot who has sick history or used to fail in medical exam and pilot who assigned more than 90 flight hours has more probability for being sick in the future period. This research also explores the other association rules that help the airline managers find the characteristics of the pilot which are going to be absent due to sickness in next period and determines the number of reserved crews should be prepared by airlines to avoid the disruption.
Keywords - Forecasting, Decision Trees, Absenteeism, Sickness, Disruption Management, Airlines, Cockpit Crew