Method and system for utilizing a trip history to predict a destination

A destination prediction system comprising a prediction unit for predicting a destination for a vehicle, the prediction unit being configured to obtain information of a current location of the vehicle and associating the current location with a current link in stored trip data comprising information of origin cluster, destination cluster and/or a series of connected links between the origin cluster and the destination cluster and probabilities for next links given a current link, to update the probabilities for next links during a trip based on a selection of a link by a user of the vehicle, and to, based on statistical properties of the current link and the stored trip data, provide a destination prediction for the vehicle including a prediction of the next link where it is possible to select between two or more alternative next links, wherein to provide the destination prediction the prediction unit is configured to handle missing data and use statistical models for prior information rather than directly use observations, adjust weights for different conditions to improve predictability, and adapt for sequentially increasing data, the prediction unit further configured to provide the destination prediction to a navigation device to provide navigation directions to a user based on the destination prediction, wherein a trajectory of the vehicle is controlled based on the navigation directions and the destination prediction.