Ticket Price Prediction Based On Deep Learning Methods

With the advent of the era of artificial intelligence,neural network technology has made great progress in various fields.Long-term and short-term memory networks are an important branch of neural networks.They not only achieve good results in the related problems of natural language processing,but also have advantages over traditional mathematical models in econometrics and statistics.Along with the development of the transportation industry and the improvement of people's living standards,more and more ordinary people have begun to use aircraft as their preferred mode of travel.For airline tickets as a special commodity,the price is easily affected by many factors,such as airlines,the number of seats remaining,the number of days in advance,the time of departure,the price of the ticket in the near time,which makes the price of the ticket change,and it is difficult for consumers.Grasping the right timing of purchase,the ticket price forecast becomes a practical research issue.The purpose of this paper is to use the deep learning method to study the change law of ticket prices,and to predict future price changes based on relevant historical change information.Construct a statistical prediction model and design a deep learning algorithm with high prediction accuracy and stable operation.This paper takes a domestic flight as a forecasting research object and predicts three changes in the price of airfare: rising,falling and unchanged.In this paper,the long-term and short-term memory network model is taken as the underlying model,and then the idea of model fusion is used to fuse it with the extreme gradient lifting tree model.Finally,the prediction effects of single model and fusion model are compared.In order to ensure the reliability of the experimental results,this paper uses the method of cross-validation as the evaluation model,multi-classification accuracy and overall category accuracy to evaluate the prediction accuracy.The experimental results show that the long-term and short-term memory network model can find out the influence and relationship between time series,and can use its unique selective memory advantage to dig deep into the inherent law of ticket price changes,but the model prediction effect is unstable.On the basis of the model,the model of the gradient lifting tree is integrated,which greatly improves the stability of the model and improves the prediction accuracy.Therefore,based on the experimental results,this paper uses the fusion model as the final ticket price prediction model.