Research And Implementation Of Airfare Prediction Technology

With the rapid development of civil aviation, more and more travellers tend to choose aviation transport for remote travel. With the development of network technology and the comprehensive promotion and usage of e-ticket, the major airlines have been using their websites to selling e-ticket, so people can retrieve information on prices of ticket from the website. When facing frequent changes on ticket prices, people are eager to know the changes of ticket variation and the proper time for booking ticket. In this article, it constructed a model based on data mining algorithms using domestic airline ticket data, with the aim at providing travelers with ticket predictions and suggestions on ticket booking.In this article, a flight of domestic subjects to used to explore underlying information from the point of view of data mining. The main work is as follows. First, collect ticket data. The article accesses to ticket price data with the assistance of the vertical search engine-HERTRIX tools, and obtains online website ticket prices through the use of HTMLParser tools. Second, outline the ticket data analysis and the pre-processing. This part includes pre-processing the data obtained, unifying data formation, storing them into database and analysinganalyzing the relationship between prices and other properties of tickets. Third, based on analysising KNN,Q-Learning and the weighted moving average time series algorithms' s basic theory, the article improves Q-Learning and the weighted moving average time series algorithms. KNN is adopted here to train the purchase decision classifier and give a suggestion on whether to buy or not, and the airfare prediction model based on improving Q-learning algorithm obtain predicted prices by training matrix Q with historical data. Finally, based on the improved and weighted moving average time series analysis forecasting model, the article studies the price prediction in two cases: less than one week and more than one week. According to the difference of forecast time with the current time, forecast prices are rendered to users. Fourth, it is the subjective Bayes algorithm integrated learning model. The article uses Bayesian reasoning techniques to integrate the three ticket price forecast model predictions to get the integrated ticket predict price and the final purchase decision. Fifth, the data acquisition price prediction techniques and integrated algorithm are combined to design a Flight price forecast prototype system.This article uses flight ticket data of CA1304 9336(Shenzhen – Beijing). KNN, Q-learning, the time sequence integration algorithms and subjective Bayes algorithm are used to predict prices, and simulation results have shown that subjective Bayes integrated algorithm is better than the other three algorithms in terms of expenditure saving.