Research On Location Prediction And Group Discovery Method Used In The Slugging Form Of Travel Mode

With the rapid development of China's economy,the number of motor vehicles has increased obviously.As a result,urban traffic congestion,environmental pollution and energy consumption has become increasingly severe.The ridesharing mode is getting more attention for its advantages of relieving traffic jams and boosting road carrying capacity.How to dig out the potential ridesharing groups from the travel information of amounts of travelers,therefore,becomes a hot issue in the current study of it.By the thorough research towards ridesharing group discovery,the method which is under the Slugging form of the travel mode and uses location prediction techniques to discover potential ridesharing group was put forward.Firstly,the stay point extraction algorithm based on conditional constraint was used to recognize travel stay information,and the stay point clustering algorithm was used to deal with the abnormal stay point,extracting the stay point trajectory,constructing the positional trajectory model,restoring the travelers' activity in the real world,introducing the POI datasets at the same time to figure out the semantic information of the geographical position,mapping geospatial location trajectories to semantic space,and constructing the semantic trajectory model.Then,the P-PPM(PrefixSpan-Prediction by Partial Matching)location prediction algorithm was proposed,with P-PPM creating a PPM prediction model based on the positional trajectory,using sequential pattern mining technology to mine frequent semantic patterns of the semantic trajectories,defining semantic movement rules based on frequent semantic patterns and correcting the result of location prediction based on its confidence.At last,a ridesharing group discovery algorithm based on location prediction was designed,and a Slugging form of ridesharing group discovery method based on location prediction was proposed.By analyzing the traveler's trajectory information and predicting the destinations of different travelers and then finding out the travelers who have the similar travel activities,a ridesharing group formed by recommendation.The Geolife datasets and the POI datasets covering 60% zone of Beijing were used to analyse the correlative parameter ? and minsup,selecting the optimal parameter value in predicting,at the same time,comparing the superiority of the P-PPM location prediction algorithm and the traditional location prediction algorithm,and applying different location prediction algorithms to the ridesharing group discovery based on location prediction,effects were made to potential ridesharing groups.The experimental results showed the P-PPM location prediction algorithm is superior to the common Markov location prediction algorithm and PPM location prediction algorithm in terms of prediction accuracy and prediction coverage.The ridesharing research method based on the P-PPM location prediction model can identify potential ridesharing groups better.