Resident Travel Characteristics Analysis And Recommendation Based On Mobile Signaling Data

With the popularity of smart phones,mobile signaling data containing geographic information can track the residents' travel trajectory very well,and these trajectory information can be further utilized to support the social networks application,intelligent transportation,smart cities,and so on.Obviously,it is of great importance to study the user-trajectory-data analysis based on mobile signaling data.On the other hand,the residents' travel often exhibits certain purposes or interests,therefore,the residents' point of interests travel recommendation is an interesting problem with increasing social network applications and the smart map evolution.Motivated by both the theoretical and the practical potentials of exploiting the user's location information delivered by the smart phone signaling data,this thesis focus on how to combine various urban points of interests and the urban residents' spatial-temporal travel distribution characteristics in order to render some travel recommendations.On the basis of the characteristics analysis of the residents' travel in urban area that is abstracted from the telecommunication signaling data and the point of interest distribution supplied by the GIS service provider,this thesis attempts to propose a preliminary technical framework to realize the resident travel recommendation system.Firstly,a DBSCAN based algorithm is realized to extract the stay-point information from the smart phone signaling data.By carefully removing the flaws,such as the inaccurate localization and the location drift in the raw mobile signaling data,the extracted stay-point information will provide us an overall picture about the residents within a specific urban area.Secondly,on the basis of the existing traffic region particitioning methods,this thesis proposes a GMM clustering based traffic cell partitioning method algorithm,wherein we may adjust the traffic cell partitioning according to the contour index to make the partitioning results more comply with the definition of traffic cell.Finally,we focus on the analysis of the inherent relationship among the resident travel distribution,the occupation and residency location as well as the points of interest.An algorithm is designed to extract the occupation and residency places of the residents from the mobile signaling data.Then a heat map is drawn to show the relationship between stay-point distribution and the points of interest in the city.Then based on the spatial-temporal distribution characteristics,this thesis proposes an algorithm to recommend the possible points of interest to fulfill residuents specified travel requirements.And the preliminary analysis and results in this thesis confirms us that,on the basis of the mobile signaling data,by analyzing the inherent relationship between the residents travel and the point of interests in city,we may find the travel point of interest characteristics,which may be utilized to provide useful point of interest recommendation for the residents travel at different time and different starting point.Although the analysis is still very preliminiary,the thesis might help some relevant research in the future.