Focasting Ticket Purchase Timing Based On Machine Learning

The global air traffic has increased dramatically in recent years,and most passengers would like to purchase tickets at a lower price.However,most airlines have their own Airline Revenue Management Systems(ARMS),which causes the prices to fluctuate frequently over time and the passengers can not make cost-minimizing decisions.If the trend of the price can be well predicted,passengers will have more opportunities to purchase cheaper tickets.So it is of great importance to study the trend of ticket price and giving ticket purchase time advice for passengers.This paper presents traditional machine learning models and deep learning models for predicting the price trends and helping passengers to capture a lower price.The traditional machine learning methods of ticket purchase timing prediction obtain the key factors affecting the ticket price through data analysis.After data analysis,the feature set is established by feature engineering,which could be the input of random forest methods and LightGBM algorithm.Attribution analysis of the features,and the importance of the features are also analyzed.The deep learning methods of ticket purchase timing prediction takes spatial and temporal factors into consideration.The temporal factor is modeled by long short-term memory networks.The spatial factor is modeled by attention mechanism.Finally,the multi-tasking model is implemented through the gated network attention mechanism to predict the trend of ticket prices of different time and flights.Finally,throught the AUC evaluation and money-saving quota evaluation,verified the effectiveness and reliability of the ticket purchase prediction model based on traditional machine learning and deep learning methods,thus helping passengers save more money in the process of purchasing a ticket.