Comparison of Different Methods for Next Location Prediction

Next location prediction anticipates a person’s movement based on the history of previous sojourns. It is useful for proactive actions taken to assist the person in an ubiquitous environment. This paper evaluates next location prediction methods: dynamic Bayesian network, multi-layer perceptron, Elman net, Markov predictor, and state predictor. For the Markov and state predictor we use additionally an optimization, the confidence counter. The criterions for the comparison are the prediction accuracy, the quantity of useful predictions, the stability, the learning, the relearning, the memory and computing costs, the modelling costs, the expandability, and the ability to predict the time of entering the next location. For evaluation we use the same benchmarks containing movement sequences of real persons within an office building.