Study On Optimal Air Ticket Purchase Timing

With the development of China's economy and aviation industry,more and more people choose aircraft as the means of transportation for long-distance travel.As an enterprise,in order to maximize its profit,airline companies will use dynamic pricing strategy to set the price of air tickets when selling air tickets.Passengers who choose the plane as a means of transportation want to be able to buy tickets when the sales price of the flight they choose is the lowest.However,air ticket is a kind of special commodity with timeliness and scarcity,and the price of air ticket is affected by multiple factors such as the departure time of the plane,the number of hours purchased in advance,and the airlines operating flights,so it is difficult for consumers to know the best time to buy air ticket.In this paper,the decision tree algorithm,Q learning algorithm and ensemble strategy in the machine learning are used to study the problem of air ticket purchase.Try to find the best time to buy a ticket,so as to help consumers save money when buying a ticket.This paper takes the route from Shanghai to Beijing as an example to study the problem of air ticket purchaseing.Try to find the best time for consumers to buy air tickets through machine learning method.This paper mainly studies the following three aspects: First of all,in the data acquisition stage,as the required data is real-time and dynamically changing data,this paper studies how to use the web crawler technology to obtain the data required by the research.Finally,we got the information of 350,000 flights of 10 airlines from Shanghai to Beijing between July 28,2018 and September 15,2018.Secondly,before studying the problem of purchasing air tickets,this paper conducts exploratory analysis on the data,studies the rule of change of air tickets and the possibility that using machine learning method can save the money.Finally,by using the decision tree algorithm,Q learning algorithm and ensemble strategy,the air ticket purchase decision problem is studied.The effectiveness of the model is evaluated from the aspects of decision accuracy and the proportion of saving funds.Compared with the single Q learning model and decision tree model,the fused model achieved better results under the two evaluation criteria of decision accuracy and capital saving rate.