Machine Learning for Airfare Prediction
This dataset is useful for evaluating or developing new prediction models of airfare prices.
The dataset contains 1814 flights. This data originally used in a 10-fold crossvalidation procedure to train and evaluate several state of the art machine learning models applied to predict the airfare prices.
Records are for flights from Thessaloniki (SKG) - Greece --> Stuttgart (STR) - Germany
This dataset was manually collected from Airtickets.gr
For every flight the following features were considered:
F1: Feature 1 - departure time.
F2: Feature 2 - arrival time.
F3: Feature 3 - number of free luggage (0, 1 or 2).
F4: Feature 4 - days left until departure.
F5: Feature 5 - number of intermediate stops.
F6: Feature 6 - holiday day (yes or no).
F7: Feature 7 - overnight flight (yes or no).
F8: Feature 8 - day of week.
If you use this dataset for your publications, please cite it as:
K. Tziridis, Th. Kalampokas, G.A. Papakostas, K.I. Diamantaras, "Airfare Prices Prediction Using Machine Learning Techniques", European Signal Processing Conference (EUSIPCO), 28 August 28 - 2 September, Kos, Greece, 2017.