Abstract:

Traditional air route value calculation is based on statistical passenger flow, ignoring the value information of passenger preference which would influence potential value of air route. In order to solve this problem, an air route potential value calculation model based on passenger travel intention is proposed in this paper. Firstly, the maximum likelihood estimation method is used to quantify the passenger preference in this model, and then the concept of a travel intention is introduced to classify the passengers travel behavior, and the Gibbs Sampling method is used to quantify the travel intention. Finally, the purpose of calculating the potential value of air route is realized. Experiments on passenger booking data of China civil aviation show that the similarity between the air routes value sequences in 2010 and 2011 respectively obtained by proposed model is much higher than that obtained by the statistical passenger flow method, and the mining accuracy on top 5 of highvalue air routes reaches 100%.

Key words: air transportion, passenger cabin preference, travel intention, air route potential value, maximum likelihood estimation, Gibbs Sampling

摘要：

传统的航线价值计算通常以统计客流量为主，忽略了旅客偏好对航线潜在价值的影响，因此，本文提出了一种基于旅客出行意图的航线潜在价值计算模型.该模型利用最大似然估计法对旅客舱位偏好进行量化，然后引入出行意图的概念将旅客出行行为进行细分，并利用Gibbs Sampling方法实现出行意图的求解，最终达到航线潜在价值计算的目的.在中国民航旅客订票数据集上的实验表明，本文方法获得的2010年航线价值序列与2011年航线价值序列的相似度要明显高于统计客流量的方法，且对排名前5的高价值航线的挖掘准确率可达100%.

关键词: 航空运输, 旅客舱位偏好, 出行意图, 航线潜在价值, 最大似然估计, Gibbs Sampling