Formulating Human Mobility Model in a Form of Continuous Time Markov Chain

Abstract It is possible for a person to collect mobility trail in a form of positioning data set with portable devices or smart phones. From such set of mobility trail we can construct human mobility model. A mobility trail is usually classified as stay states and moving states. Stay states on a specific location can be clustered and be regarded as state of Markov chain. Of course, transition probabilities between stay states can also be calculated. We used expectation maximization clustering technique and constructed Continuous Time Markov Chains representing human mobility model. In addition, we found micro mobility also. Micro mobility is mobility in a restricted area and was represented as subclusters inside a cluster. The identification of micro cluster in small area will raise another topic in human mobility model research.