[1I4-GS-2] Machine learning: Applied machine learning (1)

[1I4-GS-2-02] Airline Demand Prediction using Recurrent Neural Networks

〇Koudai Imanaka¹, Chiaki Sakama¹ (1. Wakayama University)

Keywords:Deep Learning, Recurrent Neural Network, LSTM

This study aims at constructing a machine learning system that predicts future demand of airline tickets based on past sales records. To this end, we use Sequence to Sequence (Seq2Seq) to learn reservation status and predict a demand for the next two months using the reservation status of the past two months in each booking class. Experimental results show that the system often predicts the number of remaining tickets that differs by less than three from the actual number of remaining tickets with more than 90% accuracy. The system successfully predicts airline demand depending on the seat class. In particular, it can predict the availability of tickets in high accuracy.