论文总字数：17967字

摘 要

随着国家对民航事业的重视以及航空领域的蓬勃发展，人们在日常旅行和商务活动时，越来越多的旅客因为飞机其方便快速、轻松舒适、安全可靠等优点选择乘坐飞机出行。更多的人把乘坐飞机出行作为首选的出差或旅游方式，然而各航空公司定价机制复杂，在没有充分信息的条件下，旅客们往往很难掌握价格变动的规律从而做出合理的购买决策。因此，如何买到低价飞机票成为旅客们首要关注的问题。

贝叶斯分类算法具有扎实的数学理论作为基础，基于贝叶斯分类算法中的贝叶斯网络模型，采用南京到深圳，时间为2019年9月1号——2019年9月30号的航班数据作为源数据，结合影响机票价格的各个因素，如是否是节假日、飞机起飞时段、航班编号等因素，对收集的数据进行分析与处理，构建贝叶斯网络，模型用python编码实现，应用此概率图形模型可对民航机票价格实现一定概率意义上正确的判断，最后得出一定条件下机票价格高低的概率，从而对需要乘飞机出行的旅客给出合理性建议。通过检验发现，此模型对机票价格的预测还是比较准确的。

通过本文的研究，得出以下两个结论：1）通过对样本数据的参数学习和结构学习，构造机票价格的贝叶斯网络模型；2）在此模型基础上，对机票价格进行预判，得出在一定条件下机票价格高低的概率，进而实现一定概率意义上的正确判断，给出合理性建议。

关键字：贝叶斯网络 机票价格 概率图模型

The application of bayesian classification algorithm in airline ticket price

Abstract

With the importance of nongovernmental aviation and the booming development of spaceflight，More people choose aircraft travel as their first choice of transportation. However, the pricing mechanism of airlines is complex. Without sufficient information, it is difficult for passengers to grasp the law of price changes and make reasonable purchase decisions. Therefore, how to buy low-cost air tickets has become the primary concern of passengers.

The bayesian classification algorithm has a strong mathematical theory asthe basics. the data of flight from Nanjing to Shenzhen from September 1,2019 to September 30, 2019 is used as Sample data set. Combined with various factors affecting the ticket price, such as whether it is a holiday, flight departure time, flight number and other factors, Analyzeand process the collected data, build a bayesian network, and implement the model with python code. The application of this probability graphic model can make a correct judgment on the price of civil aviation ticket in a certain probability sense, and finally get the probability of the price of air ticket under certain conditions, so as to give reasonable suggestions to the passengers who need to travel byair. After the inspection, This prediction is relatively accurate。

Through the research of the essay, the following two conclusions are drawn: 1) Through the sample data parameter learning and structure learning，Construct the bayesian network model of ticket price; 2) on the basis of this model, the probability of the high or low air ticket prices under certain conditions is obtained, then the correct judgment in the sense of certain probability is realized, and reasonable Suggestions are given.

Key Words: bayesian network; Ticket price; Probability graph model

目录

摘要................................................Ⅰ

Abstract............................................Ⅱ

1 绪论...............................................1

• 1. 研究背景及意义.................................................1

1.2 国内外研究现状................................................1

1.2.1 国外的研究现状.............................................2

1.2.2 国内的研究现状............................................2

1.3 贝叶斯网络的起源与发展.........................................2

1.4 贝叶斯网络的应用..............................................3

1.5 本文工作......................................................3

2 贝叶斯网络的基本理论..............................4

2.1 概率论与数理统计基础..........................................4

2.2 贝叶斯网络的构建..............................................5

2.3 贝叶斯网络的参数学习算法.....................................6

2.4 贝叶斯网络的结构学习算法....................................8

2.4.1 基于约束.................................................8

2.4.2 基于评分.................................................8

2.5贝叶斯网络推理................................................9

2.5.1 最大后验假设问题MAP......................................9

2.5.2 最大可能解释问题.........................................9

3 基于贝叶斯网络的机票价格分析..................11

3.1 机票价格模型介绍.............................................11

3.1.1 基于机票价格本身的历史值.................................11

3.1.2 基于影响机票价格的各种因素................................11

3.1.3 基于贝叶斯算法............................................11

3.2 贝叶斯网络模型................................................12

3.2.1 样本选取..................................................12

3.2.2 特征指标选取..............................................12

3.2.3 模型构造..................................................14

3.2.3.1 确定网络结构........................................14

3.2.3.2 确定网络参数........................................15

3.2.4 预测结果..................................................16

4 结束语.......................................18

4.1 工作总结.....................................................18

4.2 工作展望....................................................18

参考文献...........................................19

致谢...............................................22

请支付后下载全文，论文总字数：17967字