Predicting Local Airfare Prices with Deep Transfer Learning Technique

Airline ticket price is affected by several factors, i.e., flight distance, departure point, purchasing time, and the number of transfers. Machine Learning (ML) techniques are applied to predict these prices in the last decades. However, deep learning methods require a huge amount of data for training, and especially when the local dataset size is small, they could not learn their parameters well. Deep Transfer Learning (DTL) is developed for addressing this issue. In this study, we developed deep learning methods to predict flight prices for Turkey. We then applied DTL from a larger India flight price dataset to the local Turkish dataset. We presented the results and demonstrated the advantages of applying DTL to the local Turkish dataset.