Research And System Realization Of Ticket Purchase Decision Prediction Based On Machine Learning

With the development of the aviation industry and the improvement of people’s living standard,more and more people choose the aircraft as their way to travel.Most people want to buy their own ticket at a lower price,but Airlines always adjust the airline ticket price in real time according to the revenue management strategy.In the context of changing ticket prices,from the perspective of passengers,this paper,aiming at how to make a suitable purchase decision,this paper presents a short term prediction method of air ticket purchase decision and a combined algorithm of long term ticket price forecast through in-depth research and analysis of prediction methods.This paper studies and designs the ticket purchase decision-making tools to help passengers make the appropriate decisions.The short term prediction method of air ticket purchase decision establishes a predictive model based on the class variable rules and random forest algorithm.In the case of a shorter takeoff time,according to the analysis of the ticket price behavior this paper adds the two characteristics of the volatility value of the ticket price and the intensity of the price volatility constructs the forecasting model of the purchase decision,so as to suggest when the passenger should buy the ticket.The combination algorithm of long term ticket price forecast establishes a predictive model based on time series analysis and random forest regression algorithm.In the case of longer takeoff time,the article uses the time series algorithm to predict the ticket price initially,then,the random forest algorithm is used to optimize the above forecast results to make the price forecast more accurate.Finally,we design and implement the air ticket forecasting system based on machine learning.The system uses Python language and SQL Server database and follows the rules of software design and so on.It completes the requirements analysis,overall design and implementation of each module,the test proves the effectiveness of the prediction algorithm and the stability of the system.The experimental results show that the proposed algorithm is more reliable in predicting the purchase decision of air tickets and can better help passengers to make appropriate decisions.