Study Of Moving Object Location Prediction Algorithm Based On Markov Model

With the popularity of mobile devices and the development of positioning technology, the trajectory data generated by the moving object is explosive growth, how to dig out valuable information from the massive trace data is more and more concerned by scholars. Trajectory prediction technique is an important branch of trajectory data mining, through the trajectory prediction, can give the user to provide a better service experience, change the way of life of the user, etc.Firstly, by introducing the trajectory similarity algorithm into high-order Markov prediction model, given the location prediction algorithm based on high-order Markov model and trajectory similarity. For the ordinary high-order Markov prediction model, it has a high predictive sparsity in limited training data set, and the prediction result is affected by the order of the model, and the stability of the prediction model is poor. In this method, by introducing the trajectory similarity algorithm in the process of trajectory matching, and use the prediction result set of high-order Markov model as the prediction candidate set, greatly reduced the prediction of sparse rate and enhanced the stability of the prediction.Then, by introducing association rules into Markov prediction model, given the location prediction algorithm based on fixed-order Markov model and association rules. In the process of studying the high-order Markov prediction model, it is found that the state space expands, and the order number of the high-order Markov model is difficult to be determined, and in the experiment, the two-order Markov model is found to have good prediction accuracy. Although the simple Markov model has a smaller state space, but because of the lack of historical information, there is not a high precision of prediction accuracy. In this method, by introducing association rules into Markov prediction mode to revised prediction results. Not only solve the problem of state space expansion, still maintain low state space complexity and improve the prediction accuracy.At last, on the basis of theoretical research, design and realization of the moving object trajectory prediction analysis system. For different trajectory data in Geo Life trajectory data set, it can provide data processing, trajectory clustering and a variety of trajectory prediction methods, and can make a comparative analysis of the prediction results of different algorithms. At the same time, the system can provide a good visual interface to facilitate users to change the parameters of the experiment as well as a more intuitive observation of the clustering process and predict the results.