一种基于元路径拥塞模式挖掘的移动对象位置预测方法

路网上移动对象位置预测是许多位置相关服务的基础。目前移动对象位置预测方法没有充分考虑到轨迹数据中所蕴含的道路拥塞信息,而路网上的道路拥塞状态对移动对象的位置更新会产生巨大影响。提出基于元路径拥塞模式挖掘的方法(Meta-congestion-pattern mining,MCPM)。在离线挖掘阶段,从历史轨迹的频繁路径(元路径)的紧集中挖掘当地的拥塞模式,并对运动模式进行建模,其中采用基于k均值的聚类算法解决数据稀疏性问题。在线预测阶段根据挖掘的拥塞模式和运动模式依概率进行预测。最后通过理论分析和实验验证得出了算法的有效性,与相同条件下的精度预测(WhereNext,WN)方法相比,平均预测准确性提高了近20%,预测时间平均缩短了近50%。

Method to Position Prediction of Mobile Object Based on Meta-congestion-Pattern Mining

The location prediction of mobile objects on the road network is the basis of many location-related services. At present, the method for predicting the position of a moving object does not fully consider the road congestion information which mining from the trajectory data, and the congestion state has a great influence on the position update of the moving object. A meta-congestion-pattern mining (MCPM) method is proposed. In the offline mining stage, local congestion patterns are mining from the tightly-concentrated frequent paths (meta-paths) of the historical trajectories, and the movement patterns are modeled. A mean-based clustering algorithm is used to solve the data sparsity problem. The online forecasting stage predicts the probability based on the mining congestion model and the motion pattern. Finally, the validity of the algorithm is obtained by theoretical analysis and experimental verification. Compared with the WN method under the same conditions, the average prediction accuracy is improved by nearly 20%, and the prediction time is shortened by an average of nearly 50%.