Research And Application Of Trajectory Pattern Mining Algorithm Based On Mobile Dataset

In recent years, with the rapid development of location aware technology, location aware device price is also more and more cheap, mobile terminal carrying location aware device has been widely used in people’s daily life, which makes the acquisition of massive high precision of individual mobile data become possible. In addition, the rapid development of mobile Internet has accelerated location based service popular. Facebook, Renren, Foursquare, Digu etc, location based service provider, carrying location aware device of taxi and other mobile data sources are produced daily hundreds of millions of mobile data. In fact, the mobile data user left a certain extent reflects the user’s individual trajectory patterns. How to use the massive high-precision mobile data user left identify potential and meaningful trajectory pattern is the current trajectory pattern mining hot spots of theoretical research and is also currently a serious problem.In response to these problems, trajectory pattern mining algorithm proposed in this paper for carpool recommended applications. In trajectory pattern mining algorithm, firstly this paper uses data preprocessing clear invalid data and convert into mobile data users stay point data. Secondly, this paper uses trajectory pattern mining algorithm to extract the user’s trajectory pattern and uses the similar sub trajectory matching algorithm to find out the potential carpool routes between all users and calculate the carpool route’s similarity. In carpool recommended, this paper puts forward the the timestamp similarity concepts and formula. Then, this paper presents the concept of timestamp similarity and formula, combines carpooling route similarity and timestamp similarity two factors gives the formula for calculating user’s carpool route recommended degree.Last, this paper presents two kinds of carpooling recommended way with people’s daily riding habit.Based on MIT’s mobile data set, this paper conducted experiments and analysis. The results show that the proposed oriented carpool recommended applications trajectory pattern mining algorithm and trajectory pattern mining algorithm in the carpool recommended application method is feasible and effective, can be very good to find potential carpool relationship between users. Therefore, the subject proposed and the relevant research work is very necessary and significant, which has a certain role in promoting the development and application of mining trajectory patterns.