Improving mobility prediction performance with state based prediction method when the user departs from routine

With the rapid increase of the mobile users, mobility prediction has attracted more and more attention. For the moment, a lot of location prediction methods have shown that humans are highly predictable in their movements. However, the method of predicting locations when the user's behavior unexpectedly transits from routine to irregular pattern is undiscovered. In this paper, we propose a practical model based on State Based Prediction (SBP) method to predict the place to be visited when the user departs from the routine. First, we predict the routine level of user locations to discover the irregular places. Second, we use Markov method to predict future locations. Then, when user departs from the routine, we use SBP method to conduct prediction, meanwhile the prediction results of the other places remain consistent with the output of the Markov predictor. Experiments with the real mobile dataset show that our prediction model outperforms other basic mobility predictors, for example, the accuracy of our method can reach more than 83%, which is higher than the accuracy of 60% achieved by Lempel-Ziv (LZ) predictor.