Airlines fleet assignment prediction model for new flights using deep neural network

Abstract

Airline fleet assignment is the process of allocating different types of aircraft to different scheduled flight legs in order to reduce operating costs and increase revenue. In this research, flights data records from Egypt Air airlines was employed to build an intelligent fleet assignment model to predict the optimal fleet type for new flights. Deep neural network (DNN) and support vector machines (SVM) was used for model formulations. We evaluated the performance of models on a fleet type prediction. The research results showed that various accuracy levels of fleet type multiclass classifications were attained by the models. In terms of accuracy, the deep neural network performed better than support vector machines. Besides, airline companies can use our proposed model for fleet type prediction for new flight with desired parameter values 5, 20 and 250 for hidden layers, number of neuron and number of epochs respectively if they use the same structure for data attributes.