基于改进马尔可夫链的航线预测算法

发布时间：2019-01-15 07:08

【摘要】：在交通领域,研究分析旅客的出行目的地会产生很多商业价值。针对旅客出行目的地的不确定性造成研究困难的问题,现有方法利用熵衡量移动的不确定性来描述个体的出行特性,并同时考虑个体轨迹的时空相关性,并不能达到理想的预测精度,因此,提出了基于改进马尔可夫链的航线预测算法来对旅客的出行目的地进行预测。首先对旅客历史出行的距离分布、地点分布和时间规律特性进行了分析;然后又分析了人类移动对历史行为和当前地点的依赖性;最后将旅客的常住地特性和新航线的探索概率加入到转移矩阵的计算中,提出并实现了改进的马尔可夫链航线预测算法,进而对旅客的下一次出行进行预测。实验结果显示,该模型可以达到66.4%的平均预测精度。研究成果可以应用在航空领域的用户出行分析中,使航空公司更好地了解和预测旅客的出行,提供个性化的出行服务。
[Abstract]:In the field of transportation, research and analysis of passenger travel destination will produce a lot of commercial value. To solve the problem of uncertainty of passenger travel destination, the existing methods use entropy to measure the uncertainty of movement to describe the travel characteristics of individuals, and to consider the spatio-temporal correlation of individual trajectories at the same time. Therefore, a route prediction algorithm based on improved Markov chain is proposed to predict the travel destination of passengers. In this paper, the distance distribution, location distribution and temporal characteristics of historical travel are analyzed, and then the dependence of human mobility on historical behavior and current location is analyzed. Finally, the characteristics of the permanent residence of passengers and the exploration probability of the new route are added to the calculation of the transfer matrix, and an improved Markov chain route prediction algorithm is proposed and implemented, and then the next trip of the passengers is predicted. Experimental results show that the model can achieve an average prediction accuracy of 66.4%. The research results can be applied to user travel analysis in the field of aviation, so that airlines can better understand and predict passenger travel, and provide personalized travel services.
【作者单位】：四川大学计算机学院;
【基金】：国家自然科学基金资助项目(U333113) 四川省科技支撑计划项目(2014GZ0111)~~
【分类号】：O211.62;V35

【相似文献】